This article was generated entirely by AI without human authorship. Please read with discretion, or have an AI verify its accuracy.
AI TranslationSimplified ChineseEnglish
Key Insights
This issue summarizes 25 papers and preprints indexed from September 14–20, 2026, compiled by AI from abstracts. Topics include protein chemical editing and site-specific modification, cryo-EM and structure determination, ion channels and metal transport, neuropsychiatric drug targets, scientific AI agents and automation, and others. Coverage spans genetic code expansion, site-selective modification, cryo-EM foundation models, GPCR stabilization design, zinc transport, Slo1 inhibition, 5-HT7/TAAR1, paper-writing and X-ray agents, nanopore detection, and more. It highlights advances, applicability conditions, and preprint limitations.
This issue covers 25 papers and preprints added to my personal Zotero library between September 14 and September 20, 2026.
The summaries below were compiled by AI from abstracts, not from a full read of each paper. Publication/issue dates follow the original records and do not mean the work was published this week; preprints are labeled separately. The overviews and Highlights are meant for quick scanning—please consult the original papers for methods and conclusions.
Protein Chemical Editing and Site-Specific Modification (9 papers)
01 · Recent advances in site-specific modifications of peptides and proteins
Chemical Science · Publication/issue date: 2026
Authors: Xin-Rong Duan, Meng-Qian Zhang, Geng-Hui Feng, Yan-Mei Li
Site-specific modification aims to install functional groups at defined positions on a peptide or protein, so that a mixture of variously modified products does not get in the way of research and applications. This review organizes four strategies by their design logic: exploiting differences in the local chemical environment, attaching a label first and modifying second, driving reactions through proximity effects, and using enzymatic methods. The authors compare not only the reactions themselves but also why selectivity arises and how these techniques can be applied to chemical and biological questions. The article summarizes progress in site specificity, efficiency, and scope, and notes that the demands placed on these methods continue to grow. Its value lies in offering a framework for choosing tools based on the problem at hand, rather than claiming that any single modification method suits every protein, residue, and cellular context.
Highlight · The four site-specific modification strategies can be compared and selected by their source of selectivity and downstream use.
02 · Genetic Code Expansion: Recent Developments and Emerging Applications
Chemical Reviews · Publication/issue date: 2024-12-31
Authors: Yujia Huang, Pan Zhang, Haoyu Wang, Yan Chen, Tao Liu, Xiaozhou Luo
Genetic code expansion lets proteins incorporate noncanonical amino acids at defined positions, widening the space of research tools beyond nature's chemical repertoire. Starting from aminoacyl-tRNA synthetase–tRNA pairs, this review discusses progress in recognition efficiency, screening methods, translation system engineering, and noncanonical amino acid biosynthesis. Applications span the regulation of gene expression, protein engineering, and research related to drugs, vaccines, and gene editing. By connecting principles, tool improvements, and applications, the authors show that performance gains depend not only on any single recognition element but also on the translation environment as a whole. The article also surveys challenges and future opportunities; the medical uses discussed should be understood as research directions and individual case studies, not as evidence that the technology is equally mature across the board.
Highlight · Genetic code expansion introduces new chemical functionality into proteins by coordinating recognition elements with the translation machinery.
03 · Site-selective modification of native proteins
Trends in Chemistry · Publication/issue date: 2025-05-01
Authors: Yujun Kim, Han Bin Yi, Kyungdeok Seo, Hyun Soo Lee, Injae Shin
Site-selective modification of native proteins aims to add controllable functionality while preserving as much of the protein's integrity and native environment as possible. This article reviews strategies that improve selectivity—such as ligand guidance and tethered conjugation—as well as progress in ligand- and auxiliary-group-free modification of different residues and termini. These approaches help produce more homogeneous protein conjugates, antibody–drug conjugates among them, and broaden the range of sites that can be modified. The authors further emphasize the value of performing modifications under native conditions in living cells, so that protein function can be studied without removing the protein from its cellular context. Each strategy still has its own conditions of use, and understanding what these methods are good for requires weighing protein structure, reaction selectivity, and cell compatibility against one another.
Highlight · Site-selective modification in native environments links the preparation of homogeneous protein conjugates with functional studies in living cells.
04 · Chemical Editing of Proteins: From a Specific Residue to Functional Domains
Accounts of Chemical Research · Publication/issue date: 2026-04-13
Authors: Ziqi Gao, Jinpeng Zhang, Jinyu Wang, Jie Wang
This account describes the authors' framework for chemical editing of proteins, which combines synthetic chemistry with protein engineering to alter proteins at scales ranging from a single residue to functional domains. It is built around two complementary routes: genetic code expansion supplies new chemical building blocks, while directed evolution develops the enzyme tools that carry out the editing. The case studies cover spatiotemporal control of active sites, tuning the properties of catalytic pockets, stabilizing protein interaction interfaces, and domain-level replacement. At its core, the work brings chemical design, genetic encoding, and the selection of enzyme function into a single system and asks how to achieve more controllable functional changes. What the article presents is a set of methods and research experience—not a claim that any protein can now be reprogrammed at will.
Highlight · Combining chemical building blocks with enzyme editing tools enables protein functional regulation across scales, from residues to domains.
05 · Genetically Encoded Epoxide Warhead for Precise and Versatile Covalent Targeting of Proteins
Journal of the American Chemical Society · Publication/issue date: 2024-05-31
Authors: Jinpeng Zhang, Xia Wang, Qingjun Huang, Jinsong Ye, Jie Wang
Covalent protein binding must balance target recognition with the stability of the reactive group. The authors studied EPOY, a genetically encodable epoxide-containing tyrosine, and used it as a reactive unit introduced into protein conjugates. In the targets and site contexts tested, this unit formed site-specific covalent bonds with a variety of nucleophilic residues, expanding the range of reactivity available to existing covalent binding strategies. The researchers also applied it to a different class of designed binding proteins, exploring selective binding approaches against KRAS. The results show that a single chemical unit can react with a broad set of partners, though target localization and the local spatial environment still matter. The broad KRAS intervention value suggested in the abstract is a prospect for further development, not a demonstrated therapeutic effect.
Highlight · A genetically encodable epoxide reactive unit broadens the types of residues that covalent protein binding strategies can reach.
06 · A gasdermin-based life–death evolution system for reprogramming protease specificity
Nature Chemical Biology · Publication/issue date: 2026-01
Authors: Ziqi Gao, Tianzhen Li, Hao Ye, William Shu Ching Ngai, Huijie Wang, Peng R. Chen, Jie Wang
Changing protease substrate specificity could support protein editing and functional studies, but screening variant libraries large enough to find what you want is a major bottleneck. The authors propose a selection system that ties the desired cleavage behavior to host cell survival, enriching protease variants that meet the requirements through differences in growth. Using an existing protease model, they demonstrated efficient enrichment and went on to explore recognition of disease-related protein sequences. The methodological contribution is turning enzyme functions that are hard to measure one by one into selectable phenotypes, opening the door to larger-scale searches for specificity. Enrichment efficiency in these experiments does not guarantee high selectivity against any arbitrary target; final candidates still need independent measurements of activity, off-target cleavage, and performance under the specific conditions of use.
Highlight · Linking protease cleavage behavior to a selectable phenotype can scale up the search for evolved specificity.
Authors: Haoran Huang, Tao Yan, Chang Liu, Yuxiang Lu, Zhigang Wu, Xingchu Wang, Jie Wang
Histidine often takes part in nucleophilic reactions or metal coordination at enzyme active sites, and altering its chemical properties may open up new catalytic behavior. This study brought an unnatural histidine derivative into a genetic code expansion system and improved its incorporation into proteins by optimizing the recognition enzyme. The authors then tested this substitution in different catalytic contexts, observing an esterase reaction under specific acidic conditions and improved myoglobin catalysis under oxygenated conditions. The work shows that fine chemical extension of a common catalytic residue can be a complementary way to tune enzyme performance. The effects depend on the specific reaction and protein environment, however, and improvements in a handful of systems do not mean that every histidine-containing active site will benefit equally.
Highlight · Expanding the chemical forms of catalytic histidine can shift an enzyme's reactivity and condition tolerance in specific systems.
08 · Intracellular protein editing enables incorporation of noncanonical residues in endogenous proteins
Science · Publication/issue date: 2025-05
Authors: Jenna N. Beyer, Yevgeniy V. Serebrenik, Kaitlyn Toy, Mohd. Altaf Najar, Emily Feierman, Nicole R. Raniszewski, Erica Korb, Ophir Shalem, George M. Burslem
Studying proteins in their native cellular environment calls for keeping the interference from exogenous expression and engineering to a minimum. The authors combined split-intein-mediated protein splicing, genetic code expansion, and endogenous labeling to build a platform for intracellular protein editing. The method can install chemical functionality such as epitopes, specific sequences, and noncanonical amino acids at defined positions, and it extends these operations to endogenous proteins. The abstract emphasizes that this combination allows target proteins in cells to be studied with minimal perturbation, providing a tool for tracking and regulating function in live mammalian cells. Its contribution is to link endogenous localization with site-specific editing; applicability still needs to be judged by labeling position, splicing efficiency, and the function of the edited protein—one cannot assume that every target will be unaffected.
Highlight · Endogenous labeling combined with protein splicing allows noncanonical residues to be introduced into native proteins inside cells.
09 · Protein editing using a coordinated transposition reaction
Science · Publication/issue date: 2025-04-04
Authors: Yi Hua, Nicholas E. S. Tay, Xuanjia Ye, Jeremy A. Owen, Hengyuan Liu, Robert E. Thompson, Tom W. Muir
Conventional protein semisynthesis often builds target molecules by joining fragments, but the need to refold the final product limits which systems can be tackled. The authors propose a coordinated protein transposition reaction that uses mutually orthogonal split inteins to replace internal regions, effectively cutting and pasting under native folding conditions. They demonstrated the introduction of non-coded elements in multiple systems, including protein complexes that were already folded. The advantage is less reliance on global unfolding and refolding, bringing structural contexts that were previously hard to handle within reach of semisynthesis. This work expands the toolbox for internal protein editing, but whether it transfers to other targets still needs verification of local accessibility, reaction efficiency, and retention of function.
Highlight · Replacing internal protein segments under native folding conditions broadens the range of systems accessible to semisynthesis.
Structure Determination and Cryo-EM Methods (4 papers)
10 · A comprehensive foundation model for cryo-EM image processing
Nature Methods · Publication/issue date: 2026-01
Authors: Yang Yan, Shiqi Fan, Fajie Yuan, Huaizong Shen
Cryo-EM image processing depends on specialized expertise, which limits how widely the technique is adopted across labs. The authors present Cryo-IEF, pretrained unsupervised on roughly 65 million particle images to learn transferable image representations. The model can handle tasks such as classification by structure, clustering by pose, and image quality assessment. Building on this, the researchers created CryoWizard, an automated single-particle processing pipeline that ranks particle quality using the fine-tuned model and yields high-resolution structures from samples with different properties. The abstract also reports that the pipeline helps mitigate preferred orientation problems. This work shows that a foundation model can connect multiple processing tasks, though automated performance still needs to be judged against the specific sample, data quality, and reconstruction validation.
Highlight · Large-scale pretraining on particle images provides a shared representation for classification, quality assessment, and automated reconstruction.
Authors: E. Paknia, C. Flensburg, M. L. Chodkiewicz, R. H. Fogh, P. Keller, C. Vonrhein, C. Schulze-Briese, P. M. Dominiak, G. Bourenkov, G. Bricogne, A. Chari
The study reports the crystal structure of rubredoxin from the deep-sea archaeon Pyrococcus abyssi at 0.43 Å resolution, which the authors describe as the highest-resolution protein structure known at the time. Technical and workflow improvements made it possible to observe difference electron density near bond midpoints—signals that the usual spherical independent-atom model cannot fully account for. By connecting DiSCaMB's transferable aspherical atom model with BUSTER refinement, the authors eliminated the corresponding positive difference density, supporting an interpretation based on bonding electrons. The results show that combining ultra-high-quality diffraction data with more refined scattering models can extend the object of study from atomic positions to electron distribution. Routine quantum crystallography of macromolecules remains, in the authors' view, a prospect for the future.
Highlight · A 0.43 Å structure with aspherical atom refinement enables a more accurate interpretation of electron density near protein chemical bonds.
This preprint tackles compositional and conformational heterogeneity in cryo-EM. Particle classification needs a reliable structural reference, yet the reconstruction in turn depends on accurate classification — a circular dependency that can easily destabilize optimization in complex samples. CryoDECO feeds representation priors from a pretrained foundation model into an autoencoder, so that particles are first mapped into a representation space with structural meaning before classification and reconstruction proceed. The authors report that the method can distinguish one hundred structures in a simulated mixture and characterize continuous conformational changes; real data and unpurified cell extracts are also used to demonstrate its capabilities. The work proposes computational classification as a way to lower the barrier to studying complex samples, but simulated results and performance on real samples should be kept distinct — there is no basis for claiming it already replaces every biochemical purification step.
Highlight · Foundation model priors help ease the circular dependency between particle classification and 3D reconstruction.
13 · De novo design of a fusion protein tool for GPCR research
Proceedings of the National Academy of Sciences · Publication/issue date: 2025-07-22
Authors: Kaixuan Gao, Xin Zhang, Jia Nie, Hengyu Meng, Weishe Zhang, Boxue Tian, Xiangyu Liu
Structural studies of GPCRs often call for engineered stabilization, and the inactive state is especially hard to handle. The authors propose a de novo design strategy called click fusion, in which a fusion protein region with a stable structure is designed to attach rigidly to the receptor. According to the abstract, the design improves the thermal stability of the target receptor and helps determine its structure by cryo-EM. The study also shows that among structurally similar GPCRs, only minor adjustments to the linker region are needed to transfer the fusion design. Its contribution is to move the auxiliary proteins used in structural studies from empirical screening to a designable tool, with some transferability to boot. The scope still ends at the receptors already tested: structural similarity does not guarantee the same stabilizing effect or a complete absence of impact on functional states.
Highlight · A rigidly attached de novo–designed fusion domain can help stabilize GPCRs and transfer between similar receptors.
Authors: Pin Lyu, Elena Longhin, Fatemeh Sabzian-Molaei, Fernando Montalvillo Ortega, Ping Li, Kaituo Wang, Annette Duelli, Tamim Al-Jubair, Hajira Ahmed Hotiana, Eva Ramos Becares, Tristan Croll, Gabriele Meloni, Magnus Andersson, Viktoria Bågenholm, Pontus Gourdon
Zinc is an essential element, but in excess it turns toxic, so cells must coordinate sensing with transport. This study solved the inward-open structure of a dimeric zinc-transporting P-type ATPase and combined biochemical experiments with molecular dynamics to probe its regulation. The results suggest that the N-terminal tail, which carries a metal-binding domain, can self-inhibit the ATPase core, and that this inhibition is lifted when metal levels rise. The structure also reveals a negatively charged uptake region that directly exposes a conserved binding site; a nearby histidine-rich segment may help supply ions. On this basis, the authors pull the N-terminus's sensing, regulatory, and ion-supplying roles into a single model for how transport activity matches metal load — though some of the handoff steps remain a mechanistic interpretation.
Highlight · The N-terminal tail may handle zinc sensing, self-inhibition, and ion supply at once, coordinating the ATPase's transport activity.
15 · Structural underpinnings of human Slo1 inhibition by scorpion and fungal toxins
Proceedings of the National Academy of Sciences · Publication/issue date: 2026-09-22
Authors: Gopal S. Kallure, Kamalendu Pal, Gabriel W. Prather, Sandipan Chowdhury
Slo1 channels help shape electrochemical signaling in many cell types, and different inhibitors need not act the same way. Combining single-particle cryo-EM with membrane flux and binding assays, the authors compare classic pore block with gating inhibition. A fungal small molecule occupies an internal pocket of the channel and acts mainly by restricting opening; through allosteric coupling, it also changes how fast another class of blocker dissociates. The study further observed lipids entering the same pocket, hinting that the membrane environment may compete with inhibitor binding and may also alter pore hydration and gating energetics. The results tie ligands, lipids, and channel conformation together into a framework for understanding basic gating mechanisms; the various occupancy states do not all contribute identically to opening or to ligand dissociation.
Highlight · Pore block, restricted opening, and lipid occupancy form interconnected but distinct layers of Slo1 regulation.
Authors: Dmitrii Zabelskii, Sergey Bukhdruker, Gerrit H. U. Lamm, Siarhei Bukhalovich, Mako Aoyama, Vsevolod Sudarev, Alexander Kuzmin, Mikihiro Shibata, Kota Kotayama, Hideki Kandori, Josef Wachtveitl, Ernst Bamberg, Valentin Gordeliy
Viral channelrhodopsins can let cations pass under light control, yet higher calcium concentrations inhibit them. Combining crystal structures with time-resolved absorption and infrared spectroscopy, the authors studied OLPVR1 and found a transient calcium-binding site in the intracellular region near the retinal cofactor. After photoactivation, calcium there blocks the gating rearrangement required for ion transport, preventing the channel from conducting normally, while sodium accumulates elsewhere in a local region. The study thereby links a static binding site to dynamic changes across the photocycle and explains why calcium can both pass through and inhibit. These mechanistic insights could lay groundwork for future optogenetic tools, but the abstract does not show how engineered tools perform in the nervous system.
Highlight · Calcium binding can block the key gating rearrangement after photoactivation, thereby shutting down channelrhodopsin conduction.
This review examines what 5-HT7 receptor antagonism might offer depression research. The receptor is tied to mood, circadian rhythm, and cognition, signals mainly through the Gαs/cAMP pathway, and has also been reported to couple in other ways. The authors gather evidence that selective antagonists produce antidepressant-like effects in rodent behavioral models and discuss how bioengineering platforms, human-relevant models, and CNS-targeted delivery could sharpen our mechanistic understanding. They stress that because the receptor is found both peripherally and centrally, tissue selectivity of any intervention matters just as much. The existing evidence mainly supports pursuing this direction; behavioral changes in animal tests are not the same as symptom relief in patients, and they are not yet enough to define a clinical regimen.
Highlight · 5-HT7 antagonism is a direction worth exploring for antidepressants, but translation still depends on models and delivery strategies closer to human physiology.
Authors: Sijie Huang, Heng Liu, Yujin Wu, Joao M. Braz, Divya Kranthi, Brendan W. Hall, Xinyue Zhang, Dmytro S. Radchenko, Yurii S. Moroz, John J. Irwin, Allan I. Basbaum, H. Eric Xu, William C. Wetsel, Brian K. Shoichet
TAAR1 is a receptor that tunes monoamine signaling and offers a route to antipsychotic drugs beyond the classic dopamine and serotonin receptors. Using the activated-state structure, the authors ran large-scale molecular docking and tested fifty-five top-ranked candidates experimentally, finding fourteen agonists. Further optimization yielded nanomolar candidates, cryo-EM supported their predicted binding poses, and some molecules achieved high brain exposure. Three candidates corrected prepulse inhibition deficits in relevant mouse behavioral models without the catalepsy that was assayed. The study also notes that over-optimizing docking conditions for initial hit rates can work against later affinity optimization. These findings are preclinical evidence and should not be read directly as conclusions about efficacy or safety in patients.
Highlight · The activated-state TAAR1 structure enabled discovery of new agonists with brain exposure and antipsychotic-like effects in mice.
19 · Reimagining research papers as interactive and reliable AI agents
Nature · Publication/issue date: 2026-09-16
Authors: Jiacheng Miao, Joe R. Davis, Yaohui Zhang, Jonathan K. Pritchard, James Zou
Paper2Agent turns papers, supplementary materials, data, and code into research agents that can be called up in natural language, with the goal of lowering the barrier to reusing research results. The framework parses the paper and its codebase, builds MCP tool interfaces, and boosts tool reliability by generating and running tests. Using research tools such as AlphaGenome, Scanpy, and TISSUE as examples, the paper shows that the agents can reproduce original results and answer new questions, and it also demonstrates multiple agents collaborating to prioritize candidate disease genes. Its value lies in connecting static descriptions to executable workflows, making it easier for users to get inside the original paper's methods. Success in these cases does not mean any paper can automatically become a reliable tool — the outcome still hinges on the code, data, and test coverage.
Highlight · Paper agents connect research knowledge to tested, executable tools, supporting reproduction and follow-up queries.
Authors: Zhantao Chen, Alexander N. Petsch, Aidan J. Israelski, Rajan Plumley, Lingjia Shen, Cong Wang, Cheng Peng, Yuan Ni, Arun Bansil, Sugata Chowdhury, Mingda Li, Jana B. Thayer, Vivek Thampy, Joshua J. Turner
Sample alignment at synchrotron facilities usually demands constant human judgment. This study presents an LLM-driven agent that handles X-ray sample alignment by planning actions, calling instrument tools, interpreting observations, and iterating on corrections. The team first trained and tested the agent in a virtual environment simulating a real six-circle diffractometer, then deployed the workflow at an actual beamline. The agent identified reference reflections and determined the orientation matrix, and it also adjusted to unexpected experimental conditions. The results show that structured tool interfaces can plug an LLM's planning ability into specific experimental steps. This validation targets a well-defined alignment task and does not amount to the whole beamline, or any experiment, running unattended.
Highlight · After testing in a virtual environment, the agent determined sample orientation at a real synchrotron beamline.
Authors: Kieran Didi, Danny Reidenbach, Matthew Penner, Supriya Ravichandran, Marshall Case, Mike Nichols, Erik Swanson, Alex Reis, Maggie Prescott, Yue Qian, Dongming Qian, Jingjing Yang, Weiji Li, Le Li, Daichi Shonai, Sean Gay, Bhoomika Basu Mallik, Ho Yeung Chim, Liurong Chen, Miguel Atienza Juanatey, Hubert Klein, Dominic Rieger, Phillip Schlegel, Anna U. Macintyre, Maxim Secor, Daniele Granata, Sooyoung Cha, Zhonglin Cao, Guoqing Zhou, Tomas Geffner, Xi Chen, Micha Livne, Zuobai Zhang, Tianjing Zhang, Kyle Gion, Michael M. Bronstein, Martin Steinegger, Kristine Deibler, Scott Soderling, Clara T. Schoeder, Alena Khmelinskaia, Florian Hollfelder, Christian Dallago, Emine Kucukbenli, Arash Vahdat, Pierce Ogden, Karsten Kreis
This preprint tackles the difficulty of designing binding proteins for polar, solvent-exposed surfaces and flexible small molecules by proposing a latent generative search framework. The method uses reward guidance on Proteina-Complexa during inference to generate sequences and structures together, reducing reliance on a separate inverse-folding step. Through large-scale experimental screening, the authors compare several design strategies and report that joint design yields more functional binders than the other methods tested, producing high-affinity candidates against a range of protein targets. The work also demonstrates recognition of free saccharides, including designs that distinguish blood-group antigens. These results broaden the chemical space of targets that can be explored, but the preprint's comparisons are limited to its own test set and cannot be extrapolated to arbitrary targets.
Highlight · Joint sequence-and-structure generation plus inference-time search make polar surfaces and flexible sugars tractable targets for binder design.
This preprint presents an approach in which structural information drives virus discovery. Starting from virus particles in mosquito-cell-related samples, the authors first obtain a high-resolution capsid structure, then combine automated modeling with genomic analysis to link the structural observations to a complete viral identity, forming the CryoSeekV workflow. Phylogenetic comparison suggests that this virus and related mosquito-derived viruses form a distinct clade, prompting the authors to propose a new genus-level classification. The structural and cellular work also offers clues about capsid assembly and host responses. Its main contribution is showing that structural analysis can complement discovery methods that rely on sequence references; a taxonomic proposal is not the same as formal recognition, and the observed cellular phenomena cannot be used to infer the risk of human infection.
Highlight · CryoSeekV demonstrates a discovery path that starts from virus particle structure and cross-checks it against genomic information.
This study centers on the maturation and catalytic states of the coronavirus main protease. This enzyme is typically associated with dimerization-dependent activity, yet how the monomer contributes to its own maturation remains an open question. The authors report that, in in vitro work, they observed a link between peptide binding and a shift in the monomer's catalytic state, and they propose an allosteric explanation based on biophysical and computational analysis. The conceptual point is a reminder that a protein's aggregation state, local conformation and enzymatic activity may not map onto one another in a simple one-to-one way, and must be teased apart under specific experimental conditions. What this paper offers is a mechanistic clue at the molecular level; it cannot be used to infer net effects in an infection setting, viral transmission characteristics or drug efficacy, and the related tool applications remain a research prospect proposed by the authors.
Highlight · The study argues that monomer conformation and allosteric regulation need to be part of any discussion of how the main protease matures.
Authors: Maria Ciapponi, Martina Cafiso, Sven Schkölziger, Christian Benda, Jacques Bonnet, Jürg Müller
The canonical Polycomb repressive complex cPRC1 helps maintain cell fate by restraining aberrant transcription of developmental regulatory genes. The authors solved the structure of human cPRC1 in a complete complex with a modified nucleosome and a ubiquitin-conjugating enzyme, revealing a tightly integrated interface formed by several subunits that positions the relevant enzyme on the nucleosome and supports monoubiquitination of H2A. Comparative and mutational analysis in Drosophila shows that this organization is conserved, with Polyhomeotic playing a key role in assembly and targeting. The roles of its distinct domains can be separated: one portion is required for complex assembly, while another, though not required for assembly, affects recruitment to genomic sites and efficient modification. The study links structural organization to chromatin targeting.
Highlight · The scaffold organization of cPRC1 not only positions the enzyme but also, through a division of labor among its domains, connects assembly with genomic targeting.
Getting a single nanopore to distinguish several classes of biomolecules at once remains a challenge. The authors introduced a multifunctional adapter structure into an MspA nanopore, enabling it to recognize diverse analytes including amino acids, some post-translationally modified residues, nucleotides, monosaccharides and small peptides. Combined with machine learning, the study achieved an overall identification accuracy of 98.7% on the set tested. The sensor also produces predictable event features tied to analyte type, and analyses of yeast cell extracts and glycopeptide composition demonstrate its use in complex samples. The work marks a step toward unified detection across molecular classes, but benchmark accuracy does not directly represent performance on arbitrary mixed samples, and combining it with hydrolases to expand its analytical reach remains a future direction.
Highlight · A multifunctional nanopore paired with machine learning can distinguish several classes of biomolecules on a single platform.
This issue covers 25 papers and preprints added to my personal Zotero library between September 14 and September 20, 2026.
The summaries below were compiled by AI from abstracts, not from a full read of each paper. Publication/issue dates follow the original records and do not mean the work was published this week; preprints are labeled separately. The overviews and Highlights are meant for quick scanning—please consult the original papers for methods and conclusions.
Protein Chemical Editing and Site-Specific Modification (9 papers)
01 · Recent advances in site-specific modifications of peptides and proteins
Chemical Science · Publication/issue date: 2026
Authors: Xin-Rong Duan, Meng-Qian Zhang, Geng-Hui Feng, Yan-Mei Li
Site-specific modification aims to install functional groups at defined positions on a peptide or protein, so that a mixture of variously modified products does not get in the way of research and applications. This review organizes four strategies by their design logic: exploiting differences in the local chemical environment, attaching a label first and modifying second, driving reactions through proximity effects, and using enzymatic methods. The authors compare not only the reactions themselves but also why selectivity arises and how these techniques can be applied to chemical and biological questions. The article summarizes progress in site specificity, efficiency, and scope, and notes that the demands placed on these methods continue to grow. Its value lies in offering a framework for choosing tools based on the problem at hand, rather than claiming that any single modification method suits every protein, residue, and cellular context.
Highlight · The four site-specific modification strategies can be compared and selected by their source of selectivity and downstream use.
DOI: 10.1039/D5SC09331G
02 · Genetic Code Expansion: Recent Developments and Emerging Applications
Chemical Reviews · Publication/issue date: 2024-12-31
Authors: Yujia Huang, Pan Zhang, Haoyu Wang, Yan Chen, Tao Liu, Xiaozhou Luo
Genetic code expansion lets proteins incorporate noncanonical amino acids at defined positions, widening the space of research tools beyond nature's chemical repertoire. Starting from aminoacyl-tRNA synthetase–tRNA pairs, this review discusses progress in recognition efficiency, screening methods, translation system engineering, and noncanonical amino acid biosynthesis. Applications span the regulation of gene expression, protein engineering, and research related to drugs, vaccines, and gene editing. By connecting principles, tool improvements, and applications, the authors show that performance gains depend not only on any single recognition element but also on the translation environment as a whole. The article also surveys challenges and future opportunities; the medical uses discussed should be understood as research directions and individual case studies, not as evidence that the technology is equally mature across the board.
Highlight · Genetic code expansion introduces new chemical functionality into proteins by coordinating recognition elements with the translation machinery.
DOI: 10.1021/acs.chemrev.4c00216
03 · Site-selective modification of native proteins
Trends in Chemistry · Publication/issue date: 2025-05-01
Authors: Yujun Kim, Han Bin Yi, Kyungdeok Seo, Hyun Soo Lee, Injae Shin
Site-selective modification of native proteins aims to add controllable functionality while preserving as much of the protein's integrity and native environment as possible. This article reviews strategies that improve selectivity—such as ligand guidance and tethered conjugation—as well as progress in ligand- and auxiliary-group-free modification of different residues and termini. These approaches help produce more homogeneous protein conjugates, antibody–drug conjugates among them, and broaden the range of sites that can be modified. The authors further emphasize the value of performing modifications under native conditions in living cells, so that protein function can be studied without removing the protein from its cellular context. Each strategy still has its own conditions of use, and understanding what these methods are good for requires weighing protein structure, reaction selectivity, and cell compatibility against one another.
Highlight · Site-selective modification in native environments links the preparation of homogeneous protein conjugates with functional studies in living cells.
DOI: 10.1016/j.trechm.2025.03.003
04 · Chemical Editing of Proteins: From a Specific Residue to Functional Domains
Accounts of Chemical Research · Publication/issue date: 2026-04-13
Authors: Ziqi Gao, Jinpeng Zhang, Jinyu Wang, Jie Wang
This account describes the authors' framework for chemical editing of proteins, which combines synthetic chemistry with protein engineering to alter proteins at scales ranging from a single residue to functional domains. It is built around two complementary routes: genetic code expansion supplies new chemical building blocks, while directed evolution develops the enzyme tools that carry out the editing. The case studies cover spatiotemporal control of active sites, tuning the properties of catalytic pockets, stabilizing protein interaction interfaces, and domain-level replacement. At its core, the work brings chemical design, genetic encoding, and the selection of enzyme function into a single system and asks how to achieve more controllable functional changes. What the article presents is a set of methods and research experience—not a claim that any protein can now be reprogrammed at will.
Highlight · Combining chemical building blocks with enzyme editing tools enables protein functional regulation across scales, from residues to domains.
DOI: 10.1021/acs.accounts.6c00037
05 · Genetically Encoded Epoxide Warhead for Precise and Versatile Covalent Targeting of Proteins
Journal of the American Chemical Society · Publication/issue date: 2024-05-31
Authors: Jinpeng Zhang, Xia Wang, Qingjun Huang, Jinsong Ye, Jie Wang
Covalent protein binding must balance target recognition with the stability of the reactive group. The authors studied EPOY, a genetically encodable epoxide-containing tyrosine, and used it as a reactive unit introduced into protein conjugates. In the targets and site contexts tested, this unit formed site-specific covalent bonds with a variety of nucleophilic residues, expanding the range of reactivity available to existing covalent binding strategies. The researchers also applied it to a different class of designed binding proteins, exploring selective binding approaches against KRAS. The results show that a single chemical unit can react with a broad set of partners, though target localization and the local spatial environment still matter. The broad KRAS intervention value suggested in the abstract is a prospect for further development, not a demonstrated therapeutic effect.
Highlight · A genetically encodable epoxide reactive unit broadens the types of residues that covalent protein binding strategies can reach.
DOI: 10.1021/jacs.4c03974
06 · A gasdermin-based life–death evolution system for reprogramming protease specificity
Nature Chemical Biology · Publication/issue date: 2026-01
Authors: Ziqi Gao, Tianzhen Li, Hao Ye, William Shu Ching Ngai, Huijie Wang, Peng R. Chen, Jie Wang
Changing protease substrate specificity could support protein editing and functional studies, but screening variant libraries large enough to find what you want is a major bottleneck. The authors propose a selection system that ties the desired cleavage behavior to host cell survival, enriching protease variants that meet the requirements through differences in growth. Using an existing protease model, they demonstrated efficient enrichment and went on to explore recognition of disease-related protein sequences. The methodological contribution is turning enzyme functions that are hard to measure one by one into selectable phenotypes, opening the door to larger-scale searches for specificity. Enrichment efficiency in these experiments does not guarantee high selectivity against any arbitrary target; final candidates still need independent measurements of activity, off-target cleavage, and performance under the specific conditions of use.
Highlight · Linking protease cleavage behavior to a selectable phenotype can scale up the search for evolved specificity.
DOI: 10.1038/s41589-025-02063-3
07 · Genetically encoded Nδ-vinyl histidine for the evolution of enzyme catalytic center
Nature Communications · Publication/issue date: 2024-07-08
Authors: Haoran Huang, Tao Yan, Chang Liu, Yuxiang Lu, Zhigang Wu, Xingchu Wang, Jie Wang
Histidine often takes part in nucleophilic reactions or metal coordination at enzyme active sites, and altering its chemical properties may open up new catalytic behavior. This study brought an unnatural histidine derivative into a genetic code expansion system and improved its incorporation into proteins by optimizing the recognition enzyme. The authors then tested this substitution in different catalytic contexts, observing an esterase reaction under specific acidic conditions and improved myoglobin catalysis under oxygenated conditions. The work shows that fine chemical extension of a common catalytic residue can be a complementary way to tune enzyme performance. The effects depend on the specific reaction and protein environment, however, and improvements in a handful of systems do not mean that every histidine-containing active site will benefit equally.
Highlight · Expanding the chemical forms of catalytic histidine can shift an enzyme's reactivity and condition tolerance in specific systems.
DOI: 10.1038/s41467-024-50005-9
08 · Intracellular protein editing enables incorporation of noncanonical residues in endogenous proteins
Science · Publication/issue date: 2025-05
Authors: Jenna N. Beyer, Yevgeniy V. Serebrenik, Kaitlyn Toy, Mohd. Altaf Najar, Emily Feierman, Nicole R. Raniszewski, Erica Korb, Ophir Shalem, George M. Burslem
Studying proteins in their native cellular environment calls for keeping the interference from exogenous expression and engineering to a minimum. The authors combined split-intein-mediated protein splicing, genetic code expansion, and endogenous labeling to build a platform for intracellular protein editing. The method can install chemical functionality such as epitopes, specific sequences, and noncanonical amino acids at defined positions, and it extends these operations to endogenous proteins. The abstract emphasizes that this combination allows target proteins in cells to be studied with minimal perturbation, providing a tool for tracking and regulating function in live mammalian cells. Its contribution is to link endogenous localization with site-specific editing; applicability still needs to be judged by labeling position, splicing efficiency, and the function of the edited protein—one cannot assume that every target will be unaffected.
Highlight · Endogenous labeling combined with protein splicing allows noncanonical residues to be introduced into native proteins inside cells.
DOI: 10.1126/science.adr5499
09 · Protein editing using a coordinated transposition reaction
Science · Publication/issue date: 2025-04-04
Authors: Yi Hua, Nicholas E. S. Tay, Xuanjia Ye, Jeremy A. Owen, Hengyuan Liu, Robert E. Thompson, Tom W. Muir
Conventional protein semisynthesis often builds target molecules by joining fragments, but the need to refold the final product limits which systems can be tackled. The authors propose a coordinated protein transposition reaction that uses mutually orthogonal split inteins to replace internal regions, effectively cutting and pasting under native folding conditions. They demonstrated the introduction of non-coded elements in multiple systems, including protein complexes that were already folded. The advantage is less reliance on global unfolding and refolding, bringing structural contexts that were previously hard to handle within reach of semisynthesis. This work expands the toolbox for internal protein editing, but whether it transfers to other targets still needs verification of local accessibility, reaction efficiency, and retention of function.
Highlight · Replacing internal protein segments under native folding conditions broadens the range of systems accessible to semisynthesis.
DOI: 10.1126/science.adq8540
Structure Determination and Cryo-EM Methods (4 papers)
10 · A comprehensive foundation model for cryo-EM image processing
Nature Methods · Publication/issue date: 2026-01
Authors: Yang Yan, Shiqi Fan, Fajie Yuan, Huaizong Shen
Cryo-EM image processing depends on specialized expertise, which limits how widely the technique is adopted across labs. The authors present Cryo-IEF, pretrained unsupervised on roughly 65 million particle images to learn transferable image representations. The model can handle tasks such as classification by structure, clustering by pose, and image quality assessment. Building on this, the researchers created CryoWizard, an automated single-particle processing pipeline that ranks particle quality using the fine-tuned model and yields high-resolution structures from samples with different properties. The abstract also reports that the pipeline helps mitigate preferred orientation problems. This work shows that a foundation model can connect multiple processing tasks, though automated performance still needs to be judged against the specific sample, data quality, and reconstruction validation.
Highlight · Large-scale pretraining on particle images provides a shared representation for classification, quality assessment, and automated reconstruction.
DOI: 10.1038/s41592-025-02916-8
11 · Towards routine accurate electron-density studies of biological macromolecules
Acta Crystallographica Section D: Structural Biology · Publication/issue date: 2026-09-01
Authors: E. Paknia, C. Flensburg, M. L. Chodkiewicz, R. H. Fogh, P. Keller, C. Vonrhein, C. Schulze-Briese, P. M. Dominiak, G. Bourenkov, G. Bricogne, A. Chari
The study reports the crystal structure of rubredoxin from the deep-sea archaeon Pyrococcus abyssi at 0.43 Å resolution, which the authors describe as the highest-resolution protein structure known at the time. Technical and workflow improvements made it possible to observe difference electron density near bond midpoints—signals that the usual spherical independent-atom model cannot fully account for. By connecting DiSCaMB's transferable aspherical atom model with BUSTER refinement, the authors eliminated the corresponding positive difference density, supporting an interpretation based on bonding electrons. The results show that combining ultra-high-quality diffraction data with more refined scattering models can extend the object of study from atomic positions to electron distribution. Routine quantum crystallography of macromolecules remains, in the authors' view, a prospect for the future.
Highlight · A 0.43 Å structure with aspherical atom refinement enables a more accurate interpretation of electron density near protein chemical bonds.
DOI: 10.1107/S2059798326007448
12 · CryoDECO: Deconstructing Extreme Compositional and Conformational Heterogeneity in Cryo-EM via Foundation Model Priors
Langtaosha Preprint Platform (preprint) · Publication/issue date: 2026-03-02
Authors: Yang Yan, Yanwanyu Xi, Shiqi Fan, Yifei Wang, Ziyun Tang, Fajie Yuan, Huaizong Shen
This preprint tackles compositional and conformational heterogeneity in cryo-EM. Particle classification needs a reliable structural reference, yet the reconstruction in turn depends on accurate classification — a circular dependency that can easily destabilize optimization in complex samples. CryoDECO feeds representation priors from a pretrained foundation model into an autoencoder, so that particles are first mapped into a representation space with structural meaning before classification and reconstruction proceed. The authors report that the method can distinguish one hundred structures in a simulated mixture and characterize continuous conformational changes; real data and unpurified cell extracts are also used to demonstrate its capabilities. The work proposes computational classification as a way to lower the barrier to studying complex samples, but simulated results and performance on real samples should be kept distinct — there is no basis for claiming it already replaces every biochemical purification step.
Highlight · Foundation model priors help ease the circular dependency between particle classification and 3D reconstruction.
DOI: 10.65215/LTSpreprints.2025.12.30.000075
13 · De novo design of a fusion protein tool for GPCR research
Proceedings of the National Academy of Sciences · Publication/issue date: 2025-07-22
Authors: Kaixuan Gao, Xin Zhang, Jia Nie, Hengyu Meng, Weishe Zhang, Boxue Tian, Xiangyu Liu
Structural studies of GPCRs often call for engineered stabilization, and the inactive state is especially hard to handle. The authors propose a de novo design strategy called click fusion, in which a fusion protein region with a stable structure is designed to attach rigidly to the receptor. According to the abstract, the design improves the thermal stability of the target receptor and helps determine its structure by cryo-EM. The study also shows that among structurally similar GPCRs, only minor adjustments to the linker region are needed to transfer the fusion design. Its contribution is to move the auxiliary proteins used in structural studies from empirical screening to a designable tool, with some transferability to boot. The scope still ends at the receptors already tested: structural similarity does not guarantee the same stabilizing effect or a complete absence of impact on functional states.
Highlight · A rigidly attached de novo–designed fusion domain can help stabilize GPCRs and transfer between similar receptors.
DOI: 10.1073/pnas.2422360122
Ion Channels and Metal Transport (3 papers)
14 · Structural and functional insights into a Zn2+-transporting P-type ATPase
Nature Communications · Publication/issue date: 2026-09-16
Authors: Pin Lyu, Elena Longhin, Fatemeh Sabzian-Molaei, Fernando Montalvillo Ortega, Ping Li, Kaituo Wang, Annette Duelli, Tamim Al-Jubair, Hajira Ahmed Hotiana, Eva Ramos Becares, Tristan Croll, Gabriele Meloni, Magnus Andersson, Viktoria Bågenholm, Pontus Gourdon
Zinc is an essential element, but in excess it turns toxic, so cells must coordinate sensing with transport. This study solved the inward-open structure of a dimeric zinc-transporting P-type ATPase and combined biochemical experiments with molecular dynamics to probe its regulation. The results suggest that the N-terminal tail, which carries a metal-binding domain, can self-inhibit the ATPase core, and that this inhibition is lifted when metal levels rise. The structure also reveals a negatively charged uptake region that directly exposes a conserved binding site; a nearby histidine-rich segment may help supply ions. On this basis, the authors pull the N-terminus's sensing, regulatory, and ion-supplying roles into a single model for how transport activity matches metal load — though some of the handoff steps remain a mechanistic interpretation.
Highlight · The N-terminal tail may handle zinc sensing, self-inhibition, and ion supply at once, coordinating the ATPase's transport activity.
DOI: 10.1038/s41467-026-77558-1
15 · Structural underpinnings of human Slo1 inhibition by scorpion and fungal toxins
Proceedings of the National Academy of Sciences · Publication/issue date: 2026-09-22
Authors: Gopal S. Kallure, Kamalendu Pal, Gabriel W. Prather, Sandipan Chowdhury
Slo1 channels help shape electrochemical signaling in many cell types, and different inhibitors need not act the same way. Combining single-particle cryo-EM with membrane flux and binding assays, the authors compare classic pore block with gating inhibition. A fungal small molecule occupies an internal pocket of the channel and acts mainly by restricting opening; through allosteric coupling, it also changes how fast another class of blocker dissociates. The study further observed lipids entering the same pocket, hinting that the membrane environment may compete with inhibitor binding and may also alter pore hydration and gating energetics. The results tie ligands, lipids, and channel conformation together into a framework for understanding basic gating mechanisms; the various occupancy states do not all contribute identically to opening or to ligand dissociation.
Highlight · Pore block, restricted opening, and lipid occupancy form interconnected but distinct layers of Slo1 regulation.
DOI: 10.1073/pnas.2606537123
16 · Molecular mechanism of calcium inhibition in viral channelrhodopsins
Nature Communications · Publication/issue date: 2026-09-16
Authors: Dmitrii Zabelskii, Sergey Bukhdruker, Gerrit H. U. Lamm, Siarhei Bukhalovich, Mako Aoyama, Vsevolod Sudarev, Alexander Kuzmin, Mikihiro Shibata, Kota Kotayama, Hideki Kandori, Josef Wachtveitl, Ernst Bamberg, Valentin Gordeliy
Viral channelrhodopsins can let cations pass under light control, yet higher calcium concentrations inhibit them. Combining crystal structures with time-resolved absorption and infrared spectroscopy, the authors studied OLPVR1 and found a transient calcium-binding site in the intracellular region near the retinal cofactor. After photoactivation, calcium there blocks the gating rearrangement required for ion transport, preventing the channel from conducting normally, while sodium accumulates elsewhere in a local region. The study thereby links a static binding site to dynamic changes across the photocycle and explains why calcium can both pass through and inhibit. These mechanistic insights could lay groundwork for future optogenetic tools, but the abstract does not show how engineered tools perform in the nervous system.
Highlight · Calcium binding can block the key gating rearrangement after photoactivation, thereby shutting down channelrhodopsin conduction.
DOI: 10.1038/s41467-026-77716-5
Neuropsychiatric Drug Targets (2 papers)
17 · Potential role of 5-HT7 receptor antagonists in the treatment of depression: Pharmacological and bioengineering perspectives
Progress in Neuro-Psychopharmacology and Biological Psychiatry · Publication/issue date: 2026-08-30
Authors: Arghya Kusum Dhar, Ruchi Keswani, Kuttiappan Anitha, Gauri Pathak, Aryaa Nigade, Shvetank Bhatt
This review examines what 5-HT7 receptor antagonism might offer depression research. The receptor is tied to mood, circadian rhythm, and cognition, signals mainly through the Gαs/cAMP pathway, and has also been reported to couple in other ways. The authors gather evidence that selective antagonists produce antidepressant-like effects in rodent behavioral models and discuss how bioengineering platforms, human-relevant models, and CNS-targeted delivery could sharpen our mechanistic understanding. They stress that because the receptor is found both peripherally and centrally, tissue selectivity of any intervention matters just as much. The existing evidence mainly supports pursuing this direction; behavioral changes in animal tests are not the same as symptom relief in patients, and they are not yet enough to define a clinical regimen.
Highlight · 5-HT7 antagonism is a direction worth exploring for antidepressants, but translation still depends on models and delivery strategies closer to human physiology.
DOI: 10.1016/j.pnpbp.2026.111915
18 · Structure based discovery of antipsychotic-like TAAR1 agonists
Nature Communications · Publication/issue date: 2026-09-17
Authors: Sijie Huang, Heng Liu, Yujin Wu, Joao M. Braz, Divya Kranthi, Brendan W. Hall, Xinyue Zhang, Dmytro S. Radchenko, Yurii S. Moroz, John J. Irwin, Allan I. Basbaum, H. Eric Xu, William C. Wetsel, Brian K. Shoichet
TAAR1 is a receptor that tunes monoamine signaling and offers a route to antipsychotic drugs beyond the classic dopamine and serotonin receptors. Using the activated-state structure, the authors ran large-scale molecular docking and tested fifty-five top-ranked candidates experimentally, finding fourteen agonists. Further optimization yielded nanomolar candidates, cryo-EM supported their predicted binding poses, and some molecules achieved high brain exposure. Three candidates corrected prepulse inhibition deficits in relevant mouse behavioral models without the catalepsy that was assayed. The study also notes that over-optimizing docking conditions for initial hit rates can work against later affinity optimization. These findings are preclinical evidence and should not be read directly as conclusions about efficacy or safety in patients.
Highlight · The activated-state TAAR1 structure enabled discovery of new agonists with brain exposure and antipsychotic-like effects in mice.
DOI: 10.1038/s41467-026-77484-2
AI Research Agents and Automation (2 papers)
19 · Reimagining research papers as interactive and reliable AI agents
Nature · Publication/issue date: 2026-09-16
Authors: Jiacheng Miao, Joe R. Davis, Yaohui Zhang, Jonathan K. Pritchard, James Zou
Paper2Agent turns papers, supplementary materials, data, and code into research agents that can be called up in natural language, with the goal of lowering the barrier to reusing research results. The framework parses the paper and its codebase, builds MCP tool interfaces, and boosts tool reliability by generating and running tests. Using research tools such as AlphaGenome, Scanpy, and TISSUE as examples, the paper shows that the agents can reproduce original results and answer new questions, and it also demonstrates multiple agents collaborating to prioritize candidate disease genes. Its value lies in connecting static descriptions to executable workflows, making it easier for users to get inside the original paper's methods. Success in these cases does not mean any paper can automatically become a reliable tool — the outcome still hinges on the code, data, and test coverage.
Highlight · Paper agents connect research knowledge to tested, executable tools, supporting reproduction and follow-up queries.
DOI: 10.1038/s41586-026-11044-y
20 · An agentic artificially intelligent X-ray scientist
Nature Machine Intelligence · Publication/issue date: 2026-07
Authors: Zhantao Chen, Alexander N. Petsch, Aidan J. Israelski, Rajan Plumley, Lingjia Shen, Cong Wang, Cheng Peng, Yuan Ni, Arun Bansil, Sugata Chowdhury, Mingda Li, Jana B. Thayer, Vivek Thampy, Joshua J. Turner
Sample alignment at synchrotron facilities usually demands constant human judgment. This study presents an LLM-driven agent that handles X-ray sample alignment by planning actions, calling instrument tools, interpreting observations, and iterating on corrections. The team first trained and tested the agent in a virtual environment simulating a real six-circle diffractometer, then deployed the workflow at an actual beamline. The agent identified reference reflections and determined the orientation matrix, and it also adjusted to unexpected experimental conditions. The results show that structured tool interfaces can plug an LLM's planning ability into specific experimental steps. This validation targets a well-defined alignment task and does not amount to the whole beamline, or any experiment, running unattended.
Highlight · After testing in a virtual environment, the agent determined sample orientation at a real synchrotron beamline.
DOI: 10.1038/s42256-026-01261-5
Other Literature (5 papers)
21 · Latent generative search unlocks de novo design of untapped biomolecular interactions at scale
bioRxiv (preprint) · Publication/issue date: 2026-09-18
Authors: Kieran Didi, Danny Reidenbach, Matthew Penner, Supriya Ravichandran, Marshall Case, Mike Nichols, Erik Swanson, Alex Reis, Maggie Prescott, Yue Qian, Dongming Qian, Jingjing Yang, Weiji Li, Le Li, Daichi Shonai, Sean Gay, Bhoomika Basu Mallik, Ho Yeung Chim, Liurong Chen, Miguel Atienza Juanatey, Hubert Klein, Dominic Rieger, Phillip Schlegel, Anna U. Macintyre, Maxim Secor, Daniele Granata, Sooyoung Cha, Zhonglin Cao, Guoqing Zhou, Tomas Geffner, Xi Chen, Micha Livne, Zuobai Zhang, Tianjing Zhang, Kyle Gion, Michael M. Bronstein, Martin Steinegger, Kristine Deibler, Scott Soderling, Clara T. Schoeder, Alena Khmelinskaia, Florian Hollfelder, Christian Dallago, Emine Kucukbenli, Arash Vahdat, Pierce Ogden, Karsten Kreis
This preprint tackles the difficulty of designing binding proteins for polar, solvent-exposed surfaces and flexible small molecules by proposing a latent generative search framework. The method uses reward guidance on Proteina-Complexa during inference to generate sequences and structures together, reducing reliance on a separate inverse-folding step. Through large-scale experimental screening, the authors compare several design strategies and report that joint design yields more functional binders than the other methods tested, producing high-affinity candidates against a range of protein targets. The work also demonstrates recognition of free saccharides, including designs that distinguish blood-group antigens. These results broaden the chemical space of targets that can be explored, but the preprint's comparisons are limited to its own test set and cannot be extrapolated to arbitrary targets.
Highlight · Joint sequence-and-structure generation plus inference-time search make polar surfaces and flexible sugars tractable targets for binder design.
DOI: 10.64898/2026.09.12.751118
22 · CryoSeekV: Discovery of a novel virus AaTLV-IMCAS from Aedes albopictus cells by cryogenic electron microscopy
Langtaosha Preprint Platform (preprint) · Publication/issue date: 2026/09/19
Authors: Hao Qu, Linjie Li, Xiaoming Li, Yunhan Xi, Ruirui Yang, Xiao Qu, Qiuyao Jiang, Peiyi Wang, Shihua Li, George Fu Gao
This preprint presents an approach in which structural information drives virus discovery. Starting from virus particles in mosquito-cell-related samples, the authors first obtain a high-resolution capsid structure, then combine automated modeling with genomic analysis to link the structural observations to a complete viral identity, forming the CryoSeekV workflow. Phylogenetic comparison suggests that this virus and related mosquito-derived viruses form a distinct clade, prompting the authors to propose a new genus-level classification. The structural and cellular work also offers clues about capsid assembly and host responses. Its main contribution is showing that structural analysis can complement discovery methods that rely on sequence references; a taxonomic proposal is not the same as formal recognition, and the observed cellular phenomena cannot be used to infer the risk of human infection.
Highlight · CryoSeekV demonstrates a discovery path that starts from virus particle structure and cross-checks it against genomic information.
DOI: 10.65215/LTSpreprints.2026.09.19.000346
23 · Discovery of D-peptides that allosterically enhance SARS-CoV-2 3CLpro activity by stabilizing its monomeric state
Protein Science · Publication/issue date: 2026
Authors: Laiyi Feng, Xinliao Ling, Weijie Bian, Changsheng Zhang, Luhua Lai
This study centers on the maturation and catalytic states of the coronavirus main protease. This enzyme is typically associated with dimerization-dependent activity, yet how the monomer contributes to its own maturation remains an open question. The authors report that, in in vitro work, they observed a link between peptide binding and a shift in the monomer's catalytic state, and they propose an allosteric explanation based on biophysical and computational analysis. The conceptual point is a reminder that a protein's aggregation state, local conformation and enzymatic activity may not map onto one another in a simple one-to-one way, and must be teased apart under specific experimental conditions. What this paper offers is a mechanistic clue at the molecular level; it cannot be used to infer net effects in an infection setting, viral transmission characteristics or drug efficacy, and the related tool applications remain a research prospect proposed by the authors.
Highlight · The study argues that monomer conformation and allosteric regulation need to be part of any discussion of how the main protease matures.
DOI: 10.1002/pro.70788
24 · Cryo-EM structure, enzymatic activity and genome targeting of canonical PRC1
Nature Structural & Molecular Biology · Publication/issue date: 2026-09-14
Authors: Maria Ciapponi, Martina Cafiso, Sven Schkölziger, Christian Benda, Jacques Bonnet, Jürg Müller
The canonical Polycomb repressive complex cPRC1 helps maintain cell fate by restraining aberrant transcription of developmental regulatory genes. The authors solved the structure of human cPRC1 in a complete complex with a modified nucleosome and a ubiquitin-conjugating enzyme, revealing a tightly integrated interface formed by several subunits that positions the relevant enzyme on the nucleosome and supports monoubiquitination of H2A. Comparative and mutational analysis in Drosophila shows that this organization is conserved, with Polyhomeotic playing a key role in assembly and targeting. The roles of its distinct domains can be separated: one portion is required for complex assembly, while another, though not required for assembly, affects recruitment to genomic sites and efficient modification. The study links structural organization to chromatin targeting.
Highlight · The scaffold organization of cPRC1 not only positions the enzyme but also, through a division of labor among its domains, connects assembly with genomic targeting.
DOI: 10.1038/s41594-026-01885-6
25 · An engineered nanopore identifies saccharides, amino acids, peptides and ribonucleotides
Nature Biotechnology · Publication/issue date: 2026-09-14
Authors: Lang Yao, Zixuan Wang, Jialu Chen, Wen Sun, Kefan Wang, Yunqi Xiao, Hanhan Zhang, Wenzheng Li, Yifan Wang, Lulu Zhao, Xinyi Dai, Lu Qian, Panke Zhang, Shuo Huang
Getting a single nanopore to distinguish several classes of biomolecules at once remains a challenge. The authors introduced a multifunctional adapter structure into an MspA nanopore, enabling it to recognize diverse analytes including amino acids, some post-translationally modified residues, nucleotides, monosaccharides and small peptides. Combined with machine learning, the study achieved an overall identification accuracy of 98.7% on the set tested. The sensor also produces predictable event features tied to analyte type, and analyses of yeast cell extracts and glycopeptide composition demonstrate its use in complex samples. The work marks a step toward unified detection across molecular classes, but benchmark accuracy does not directly represent performance on arbitrary mixed samples, and combining it with hydrolases to expand its analytical reach remains a future direction.
Highlight · A multifunctional nanopore paired with machine learning can distinguish several classes of biomolecules on a single platform.
DOI: 10.1038/s41587-026-03308-9