Harnessing protein folding neural networks for peptide-protein docking

NCPepFold: Accurate Prediction of Noncanonical Cyclic Peptide Structures via Cyclization Optimization with Multigranular Representation

Artificial intelligence-based peptide structure prediction methods have revolutionized biomolecular science. However, restricting predictions to peptides composed solely of 20 natural amino acids significantly limits their practical application; as such, peptides often demonstrate poor stability under physiological conditions. Here, we present NCPepFold, a computational approach that can utilize a specific cyclic position matrix to directly predict the structure of cyclic peptides with noncanonical amino acids. By integrating multigranularity information at the residual and atomic level, along with fine-tuning techniques, NCPepFold significantly improves prediction accuracy, with the average peptide root-mean-square deviation (RMSD) for cyclic peptides being 1.640 Å. In summary, this is a novel deep learning model designed specifically for cyclic peptides with noncanonical amino acids, offering great potential for peptide drug design and advancing biomedical research.

Recent progress and future challenges in structure-based protein-protein interaction prediction

Protein-protein interactions (PPIs) play a fundamental role in cellular processes, and understanding these interactions is crucial for advances in both basic biological science and biomedical applications. This review presents an overview of recent progress in computational methods for modeling protein complexes and predicting PPIs based on 3D structures, focusing on the transformative role of artificial intelligence-based approaches. We further discuss the expanding biomedical applications of PPI research, including the elucidation of disease mechanisms, drug discovery, and therapeutic design. Despite these advances, significant challenges remain in predicting host-pathogen interactions, interactions between intrinsically disordered regions, and interactions related to immune responses. These challenges are worthwhile for future explorations and represent the frontier of research in this field.

Predicting the structures of cyclic peptides containing unnatural amino acids by HighFold2

Cyclic peptides containing unnatural amino acids possess many excellent properties and have become promising candidates in drug discovery. Therefore, accurately predicting the 3D structures of cyclic peptides containing unnatural residues will significantly advance the development of cyclic peptide-based therapeutics. Although deep learning-based structural prediction models have made tremendous progress, these models still cannot predict the structures of cyclic peptides containing unnatural amino acids. To address this gap, we introduce a novel model, HighFold2, built upon the AlphaFold-Multimer framework. HighFold2 first extends the pre-defined rigid groups and their initial atomic coordinates from natural amino acids to unnatural amino acids, thus enabling structural prediction for these residues. Then, it incorporates an additional neural network to characterize the atom-level features of peptides, allowing for multi-scale modeling of peptide molecules while enabling the distinction between various unnatural amino acids. Besides, HighFold2 constructs a relative position encoding matrix for cyclic peptides based on different cyclization constraints. Except for training using spatial structures with unnatural amino acids, HighFold2 also parameterizes the unnatural amino acids to relax the predicted structure by energy minimization for clash elimination. Extensive empirical experiments demonstrate that HighFold2 can accurately predict the 3D structures of cyclic peptide monomers containing unnatural amino acids and their complexes with proteins, with the median RMSD for Cα reaching 1.891 Å. All these results indicate the effectiveness of HighFold2, representing a significant advancement in cyclic peptide-based drug discovery.

BRD9 functions as a methylarginine reader to regulate AKT-EZH2 signaling

Recognition of methylarginine marks by effector proteins ("readers") is a critical link between arginine methylation and various cellular processes. Recently, we identified methylation of AKT1 at arginine-391 (R391), but the reader for this methylation has yet to be characterized. Here, we show that bromodomain-containing protein 9 (BRD9), a reader of acetylated lysine, unexpectedly recognizes methylated R391 of AKT1 through an aromatic cage in its bromodomain. Disrupting the methylarginine reader function of BRD9 suppresses AKT activation and tumorigenesis. RNA sequencing data show that BRD9 and AKT coregulate a hallmark transcriptional program in part through enhancer of zeste homolog 2 (EZH2)-mediated methylation of histone-3 lysine-27. We also find that inhibitors of BRD9 and EZH2 display synergistic effects on suppression of cell proliferation and tumor growth. Collectively, our study reveals a previously unknown function of BRD9 and a potential therapeutic strategy for cancer treatment by combining BRD9 and EZH2 inhibitors.

Sensory plasticity caused by up-down regulation encodes the information of short-term learning and memory

Learning and memory are essential for animals' well-being and survival. The underlying mechanisms are a major task of neuroscience studies. In this study, we identified a circuit consisting of ASER, RIC, RIS, and AIY, is required for short-term salt chemotaxis learning (SCL) in C. elegans. ASER NaCl-sensation possesses are remodeled by salt/food-deprivation pared conditioning. RIC integrates the sensory information of NaCl and food availability. It excites ASER and inhibits AIY by tyramine/TYRA-2 and octopamine/OCTR-1 signaling pathways, respectively. By the salt conditioning, RIC NaCl calcium response to NaCl is depressed, thus, the RIC excitation of ASER and inhibition of AIY are suppressed. ASER excites RIS by FLP-14/FRPR-10 signaling. RIS inhibits ASER via PDF-2/PDFR-1 signaling in negative feedback. ASER sensory plasticity caused by RIC plasticity and RIS negative feedback are required for both learning and memory recall. Thus, the sensation plasticity encodes the information of the short-term SCL that facilitates animal adaptation to dynamic environments.

Unraveling the hidden complexity of cancer through long-read sequencing

Cancer is fundamentally a disease of the genome, characterized by extensive genomic, transcriptomic, and epigenomic alterations. Most current studies predominantly use short-read sequencing, gene panels, or microarrays to explore these alterations; however, these technologies can systematically miss or misrepresent certain types of alterations, especially structural variants, complex rearrangements, and alterations within repetitive regions. Long-read sequencing is rapidly emerging as a transformative technology for cancer research by providing a comprehensive view across the genome, transcriptome, and epigenome, including the ability to detect alterations that previous technologies have overlooked. In this Perspective, we explore the current applications of long-read sequencing for both germline and somatic cancer analysis. We provide an overview of the computational methodologies tailored to long-read data and highlight key discoveries and resources within cancer genomics that were previously inaccessible with prior technologies. We also address future opportunities and persistent challenges, including the experimental and computational requirements needed to scale to larger sample sizes, the hurdles in sequencing and analyzing complex cancer genomes, and opportunities for leveraging machine learning and artificial intelligence technologies for cancer informatics. We further discuss how the telomere-to-telomere genome and the emerging human pangenome could enhance the resolution of cancer genome analysis, potentially revolutionizing early detection and disease monitoring in patients. Finally, we outline strategies for transitioning long-read sequencing from research applications to routine clinical practice.

ATOMICA: Learning Universal Representations of Intermolecular Interactions

Molecular interactions underlie nearly all biological processes, but most machine learning models treat molecules in isolation or specialize in a single type of interaction, such as protein-ligand or protein-protein binding. This siloed approach prevents generalization across biomolecular classes and limits the ability to model interaction interfaces systematically. We introduce ATOMICA, a geometric deep learning model that learns atomic-scale representations of intermolecular interfaces across diverse biomolecular modalities, including small molecules, metal ions, amino acids, and nucleic acids. ATOMICA uses a self-supervised denoising and masking objective to train on 2,037,972 interaction complexes and generate hierarchical embeddings at the levels of atoms, chemical blocks, and molecular interfaces. The model generalizes across molecular classes and recovers shared physicochemical features without supervision. Its latent space captures compositional and chemical similarities across interaction types and follows scaling laws that improve representation quality with increasing biomolecular data modalities. We apply ATOMICA to construct five modality-specific interfaceome networks, termed ATOMICAN et s, which connect proteins based on interaction similarity with ions, small molecules, nucleic acids, lipids, and proteins. These networks identify disease pathways across 27 conditions and predict disease-associated proteins in autoimmune neuropathies and lymphoma. Finally, we use ATOMICA to annotate the dark proteome-proteins lacking known structure or function-by predicting 2,646 previously uncharacterized ligand-binding sites. These include putative zinc finger motifs and transmembrane cytochrome subunits, demonstrating that ATOMICA enables systematic annotation of molecular interactions across the proteome.

Longitudinal trajectories of brain development from infancy to school age and their relationship to literacy development

Reading is one of the most complex skills that we utilize daily, and it involves the early development and interaction of various lower-level subskills, including phonological processing and oral language. These subskills recruit brain structures, which begin to develop long before the skill manifests and exhibit rapid development during infancy. However, how longitudinal trajectories of early brain development in these structures support long-term acquisition of literacy subskills and subsequent reading is unclear. Children underwent structural and diffusion MRI scanning at multiple timepoints between infancy and second grade and were tested for literacy subskills in preschool and decoding and word reading in early elementary school. We developed and implemented a reproducible pipeline to generate longitudinal trajectories of early brain development to examine associations between these trajectories and literacy (sub)skills. Furthermore, we examined whether familial risk of reading difficulty and children's home literacy environments, two common literacy-related covariates, influenced those trajectories. Results showed that individual differences in curve features (e.g., intercepts and slopes) for longitudinal trajectories of volumetric, surface-based, and white matter organization measures were linked directly to phonological processing and indirectly to first-grade decoding and word reading skills via phonological processing. Altogether, these findings suggest that the brain bases of phonological processing, previously identified as the strongest behavioral predictor of reading and decoding skills, may already begin to develop by birth but undergo further refinement between infancy and preschool. The present study underscores the importance of considering academic skill acquisition from the very beginning of life.

Molecular Modelling in Bioactive Peptide Discovery and Characterisation

Molecular modelling is a vital tool in the discovery and characterisation of bioactive peptides, providing insights into their structural properties and interactions with biological targets. Many models predicting bioactive peptide function or structure rely on their intrinsic properties, including the influence of amino acid composition, sequence, and chain length, which impact stability, folding, aggregation, and target interaction. Homology modelling predicts peptide structures based on known templates. Peptide-protein interactions can be explored using molecular docking techniques, but there are challenges related to the inherent flexibility of peptides, which can be addressed by more computationally intensive approaches that consider their movement over time, called molecular dynamics (MD). Virtual screening of many peptides, usually against a single target, enables rapid identification of potential bioactive peptides from large libraries, typically using docking approaches. The integration of artificial intelligence (AI) has transformed peptide discovery by leveraging large amounts of data. AlphaFold is a general protein structure prediction tool based on deep learning that has greatly improved the predictions of peptide conformations and interactions, in addition to providing estimates of model accuracy at each residue which greatly guide interpretation. Peptide function and structure prediction are being further enhanced using Protein Language Models (PLMs), which are large deep-learning-derived statistical models that learn computer representations useful to identify fundamental patterns of proteins. Recent methodological developments are discussed in the context of canonical peptides, as well as those with modifications and cyclisations. In designing potential peptide therapeutics, the main outstanding challenge for these methods is the incorporation of diverse non-canonical amino acids and cyclisations.

The Effect of Microsaccades in the Primary Visual Cortex: Increased Synchronization in the Fovea during a Two-Phase Response Modulation

Our eyes are never still. Even when we attempt to fixate, the visual gaze is never motionless, as we continuously perform miniature oculomotor movements termed as fixational eye movements. The fastest eye movements during the fixation epochs are termed microsaccades (MSs) that are leading to continual motion of the visual input, affecting mainly neurons in the fovea. Yet our vision appears to be stable. To explain this gap, previous studies suggested the existence of an extraretinal input (ERI) into the visual cortex that can account for the motion and produce visual stability. Here, we investigated the existence of an ERI to V1 fovea in macaque monkeys (male) while they performed spontaneous MSs, during fixation. We used voltage-sensitive dye imaging (VSDI) to measure and characterize at high spatiotemporal resolution the influence of MSs on neural population activity, in the foveal region of the primary visual cortex (V1). Microsaccades, performed over a blank screen, induced a two-phase response modulation: an early suppression followed by an enhancement. A correlation analysis revealed a widespread foveal increase in neural synchronization, peaking around ∼100 ms after MS onset. Next, we investigated the MS effects in the presence of a small visual stimulus and found that this modulation was different from the blank condition yet both modulations coexisted in the fovea. Finally, the VSD response to an external motion of the fixation point could not explain the MS modulation. These results support an ERI that may be involved in visual stabilization already at the level of V1.

Bias in, bias out - AlphaFold-Multimer and the structural complexity of protein interfaces

A structural understanding of protein-protein interactions is a key component of many facets of applied molecular biology research. AlphaFold-Multimer (AF-MM) provided a breakthrough in the ability to predict protein-protein interface structure. However, the available training data for this model and the resulting benchmarking and validation efforts show a bias toward interactions between more ordered regions of proteins. Here we highlight some of the successes and limitations of AF-MM and discuss available methods and future directions to enable balanced prediction of all interface types.

Heuristic energy-based cyclic peptide design

Rational computational design is crucial to the pursuit of novel drugs and therapeutic agents. Meso-scale cyclic peptides, which consist of 7-40 amino acid residues, are of particular interest due to their conformational rigidity, binding specificity, degradation resistance, and potential cell permeability. Because there are few natural cyclic peptides, de novo design involving non-canonical amino acids is a potentially useful goal. Here, we develop an efficient pipeline (CyclicChamp) for cyclic peptide design. After converting the cyclic constraint into an error function, we employ a variant of simulated annealing to search for low-energy peptide backbones while maintaining peptide closure. Compared to the previous random sampling approach, which was capable of sampling conformations of cyclic peptides of up to 14 residues, our method both greatly accelerates the computation speed for sampling conformations of small macrocycles (ca. 7 residues), and addresses the high-dimensionality challenge that large macrocycle designs often encounter. As a result, CyclicChamp makes conformational sampling tractable for 15- to 24-residue cyclic peptides, thus permitting the design of macrocycles in this size range. Microsecond-length molecular dynamics simulations on the resulting 15, 20, and 24 amino acid cyclic designs identify designs with kinetic stability. To test their thermodynamic stability, we perform additional replica exchange molecular dynamics simulations and generate free energy surfaces. Three 15-residue designs, one 20-residue and one 24-residue design emerge as promising candidates.

Protein-Peptide Docking with ESMFold Language Model

Designing peptide therapeutics requires precise peptide docking, which remains a challenge. We assessed the ESMFold language model, originally designed for protein structure prediction, for its effectiveness in protein-peptide docking. Various docking strategies, including polyglycine linkers and sampling-enhancing modifications, were explored. The number of acceptable-quality models among top-ranking results is comparable to traditional methods and generally lower than AlphaFold-Multimer or Alphafold 3, though ESMFold surpasses it in some cases. The combination of result quality and computational efficiency underscores ESMFold's potential value as a component in a consensus approach for high-throughput peptide design.

Translating the Post-Mortem Brain Multi-Omics Molecular Taxonomy of Alzheimer's Dementia to Living Humans

Alzheimer's disease (AD) dementia is characterized by significant molecular and phenotypic heterogeneity, which confounds its mechanistic understanding, diagnosis, and effective treatment. In this study, we harness the most comprehensive dataset of paired ante-mortem blood omics, clinical, psychological, and post-mortem brain multi-omics data and neuroimaging to extensively characterize and translate the molecular taxonomy of AD dementia to living individuals. First, utilizing a comprehensive integration of eight complementary molecular layers from brain multi-omics data (N = 1,189), we identified three distinct molecular AD dementia subtypes exhibiting strong associations with cognitive decline, sex, psychological traits, brain morphology, and characterized by specific cellular and molecular drivers involving immune, vascular, and oligodendrocyte precursor cells. Next, in a significant translational effort, we developed predictive models to convert these advanced brain-derived molecular profiles (AD dementia pseudotimes and subtypes) into blood-, MRI- and psychological traits-based markers. The translation results underscore both the promise of these models and the opportunities for further enhancement. Our findings enhance the understanding of AD heterogeneity, underscore the value of multi-scale molecular approaches for elucidating causal mechanisms, and lay the groundwork for the development of novel therapies in living persons that target multi-level brain molecular subtypes of AD dementia.

Chromosome-level genomics and historical museum collections reveal new insights into the population structure and chromosome evolution of waterbuck

Advances in the sequencing and assembly of chromosome-level genome assemblies has enabled the study of non-model animals, providing further insights into the evolution of genomes and chromosomes. Here, we present the waterbuck ( Kobus ellipsiprymnus ) as an emerging model antelope for studying population dynamics and chromosome evolution. Antelope evolutionary history has been shaped by Robertsonian (Rb) fusions, with waterbuck also showing variation in karyotype due to two polymorphic Rb fusions. These polymorphisms are variable between and within the two recognised subspecies, the common and defassa waterbuck. To provide new insights into waterbuck evolution, we firstly assembled a chromosome-level genome assembly for the defassa subspecies using PacBio HiFi and Hi-C sequencing. We then utilised museum collections to carry out whole genome sequencing (WGS) of 24 historical waterbuck skins from both subspecies. Combined with a previous WGS dataset (n = 119), this represents the largest study of waterbuck populations to date. We found novel population structure and gene flow between waterbuck populations and regions across the genome with high genomic differentiation between the two subspecies. Several of these regions were found around the centromeres of fixed and polymorphic Rb fusions, exhibiting signatures of low recombination and local population structure. Interestingly, these regions contain genes involved in development, fertility, and recombination. Our results highlight the importance of assembling genomes to the chromosome-level, the utility and value of historical collections in sampling a wide-ranging species to uncover fine-scale population structure, and the potential impacts of Rb fusions on genomic differentiation and the recombination landscape.

Neutrophil elastase binds at the central domain of extracellular Toll-like receptor 4: AI prediction, docking, and validation in disease model

The interaction between Neutrophil Elastase (NE) and Toll-like receptor 4 (TLR4) has attracted substantial scientific attention, particularly regarding its potential role in cardiovascular diseases. Employing AlphaFold2, biomolecular docking, and MMGBSA calculation we aimed to predict their binding and validated the results through a co-immunoprecipitation study in a rat model with isoproterenol (ISO) -induced cardiac hypertrophy. Our findings strongly suggest a specific and plausible interaction between rat NE and rat TLR4, distinct from other neutrophil-derived serine proteases. Notably, AlphaFold2's precision was confirmed through cross-validation with known protein crystal structures, while Consurf analysis emphasized the evolutionary variable to conserve the rat NE - rat TLR4 binding site. HADDOCK, RosettaDock, ZDOCK, MD simulation, MMGBSA calculations, and superimposition with the stabilized structure complex all predicted strong binding between rat NE and rat TLR4. Our animal experiments revealed elevated NE and TLR4 expression in the hypertrophied myocardium following ISO infusion, with data confirming the physical interaction between NE and TLR4. Overall, this study sheds light on the intricate molecular association between NE and TLR4, underlining their potential significance in cardiovascular pathophysiology. Furthermore, it underscores AlphaFold2's reliability as a robust tool for predicting protein-protein interactions and complex structures.

The hidden bacterial microproteome

Microproteins encoded by small open reading frames comprise the "dark matter" of proteomes. Although microproteins have been detected in diverse organisms from all three domains of life, many more remain to be identified, and only a few have been functionally characterized. In this comprehensive study of intergenic small open reading frames (ismORFs, 15-70 codons) in 5,668 bacterial genomes of the family Enterobacteriaceae, we identify 67,297 clusters of ismORFs subject to purifying selection. Expression of tagged Escherichia coli microproteins is detected for 11 of the 16 tested, validating the predictions. Although the ismORFs mainly code for hydrophobic, potentially transmembrane, unstructured, or minimally structured microproteins, some globular folds, oligomeric structures, and possible interactions with proteins encoded by neighboring genes are predicted. Complete information on the predicted microprotein families, including evidence of transcription and translation, and structure predictions are available as an easily searchable resource for investigation of microprotein functions.

Comprehensive review of extraction, purification, structural characteristics, pharmacological activities, structure-activity relationship and application of seabuckthorn protein and peptides

Seabuckthorn (Hippophae rhamnoides) is an excellent plant that has the concomitant function of both medicine and foodstuff with high nutritional and health-promoting properties. As a pivotal bioactive component mainly existing in the seeds and leaves, seabuckthorn protein and its derived peptides have aroused wide attention owing to their multifaceted pharmacological activities, including anti-hypertensive, hypoglycemic, anti-obesity, anti-freeze, immunomodulatory, anti-inflammatory, sobriety, anti-oxidant and anti-neurodegenerative functions. Despite these promising attributes, the application of seabuckthorn peptides as functional food and medicines are impeded due to lack of a comprehensive understanding of pharmacological activities and intricate structure-activity relationship. Therefore, this review systematically summarizes the latest advancements in the extraction, purification, structural characteristics, pharmacological activities, digestion, absorption and transport, and application of seabuckthorn protein or peptides. Noteworthily, the structure-activity relationship is specifically delved into the hypoglycemic, anti-hypertensive, anti-obesity, anti-neurodegenerative and anti-oxidant peptides. Moreover, the shortcomings of current research and promising prospects are also highlighted. This comprehensive overview will provide a framework for future exploration and application of seabuckthorn protein or peptides in the realms of food and pharmaceuticals, offering a promising horizon for health benefits.

Heuristic energy-based cyclic peptide design

Rational computational design is crucial to the pursuit of novel drugs and therapeutic agents. Meso-scale cyclic peptides, which consist of 7-40 amino acid residues, are of particular interest due to their conformational rigidity, binding specificity, degradation resistance, and potential cell permeability. Because there are few natural cyclic peptides, de novo design involving non-canonical amino acids is a potentially useful goal. Here, we develop an efficient pipeline (CyclicChamp) for cyclic peptide design. After converting the cyclic constraint into an error function, we employ a variant of simulated annealing to search for low-energy peptide backbones while maintaining peptide closure. Compared to the previous random sampling approach, which was capable of sampling conformations of cyclic peptides of up to 14 residues, our method both greatly accelerates the computation speed for sampling conformations of small macrocycles (ca. 7 residues), and addresses the high-dimensionality challenge that large macrocycle designs often encounter. As a result, CyclicChamp makes conformational sampling tractable for 15- to 24-residue cyclic peptides, thus permitting the design of macrocycles in this size range. Microsecond-length molecular dynamics simulations on the resulting 15, 20, and 24 amino acid cyclic designs identify designs with kinetic stability. To test their thermodynamic stability, we perform additional replica exchange molecular dynamics simulations and generate free energy surfaces. Three 15-residue designs, one 20-residue and one 24-residue design emerge as promising candidates.

Trends in the research and development of peptide drug conjugates: artificial intelligence aided design

Peptide-drug conjugates (PDCs) represent an emerging class of targeted therapeutic agents that consist of small molecular drugs coupled to multifunctional peptides through cleavable or non-cleavable linkers. The principal advantage of PDCs lies in their capacity to deliver drugs to diseased tissues at increased local concentrations, thereby reducing toxicity and mitigating adverse effects by limiting damage to non-diseased tissues. Despite the increasing number of PDCs being developed for various diseases, their advancements remain relatively slow due to several development constraints, which include limited available peptides and linkers, narrow therapeutic applications, and incomplete evaluation and information platforms for PDCs. Marked by the recent Nobel Prize awarded to artificial intelligence (AI) and de novo protein design for "protein design and structure prediction," AI is playing an increasingly important role in drug discovery and development. In this review, we summarize the recent developments and limitations of PDCs, highlights the potential of AI in revolutionizing the design and evaluation of PDC.

Dynamic Modeling of Cell Cycle Arrest Through Integrated Single-Cell and Mathematical Modelling Approaches

Highly multiplexed imaging assays allow simultaneous quantification of multiple protein and phosphorylation markers, providing a static snapshots of cell types and states. Pseudo-time techniques can transform these static snapshots of unsynchronized cells into dynamic trajectories, enabling the study of dynamic processes such as development trajectories and the cell cycle. Such ordering also enables training of mathematical models on these data, but technical challenges have hitherto made it difficult to integrate multiple experimental conditions, limiting the predictive power and insights these models can generate. In this work, we propose data processing and model training approaches for integrating multiplexed, multi-condition immunofluorescence data with mathematical modelling. We devise training strategies that are applicable to datasets where cells exhibit oscillatory as well as arrested dynamics and use them to train a cell cycle model on a dataset of MCF-10A mammary epithelial exposed to cell-cycle arresting small molecules. We validate the model by investigating predicted growth factor sensitivities and responses to inhibitors of cells at different initial conditions. We anticipate that our framework will generalise to other highly multiplexed measurement techniques such as mass-cytometry, rendering larger bodies of data accessible to dynamic modelling and paving the way to deeper biological insights.

Network properties constrain natural selection on gene expression in Caenorhabditis elegans

Gene regulatory networks (GRNs) integrate genetic and environmental signals to coordinate complex phenotypes and evolve through a balance of selection and drift. Using publicly available datasets from Caenorhabditis elegans, we investigated the extent of natural selection on transcript abundance by linking population-scale variation in gene expression to fecundity, a key fitness component. While the expression of most genes covaried only weakly with fitness, which is typical for polygenic traits, we identified seven transcripts under significant directional selection. These included nhr-114 and feh-1, implicating variation in nutrient-sensing and metabolic pathways as impacting fitness. Stronger directional selection on tissue-specific and older genes highlighted the germline and nervous system as focal points of adaptive change. Network position further constrained selection on gene expression; high-connectivity genes faced stronger stabilizing and directional selection, highlighting GRN architecture as a key factor in microevolutionary dynamics. The activity of transcription factors such as zip-3, which regulates mitochondrial stress responses, emerged as targets of selection, revealing potential links between energy homeostasis and fitness. Our findings demonstrate how GRNs mediate the interplay between selection and drift, shaping microevolutionary trajectories of gene expression and phenotypic diversity.

Depolymerizing F-actin accelerates the exit from pluripotency to enhance stem cell-derived islet differentiation

In this study, we demonstrate that cytoskeletal state at the onset of directed differentiation is critical for the specification of human pluripotent stem cells (hPSCs) to all three germ layers. In particular, a polymerized actin cytoskeleton facilitates directed ectoderm differentiation, while depolymerizing F-actin promotes mesendoderm lineages. Applying this concept to a stem cell-derived islet (SC-islet) differentiation protocol, we show that depolymerizing F-actin with latrunculin A (latA) during the first 24 hours of definitive endoderm formation facilitates rapid exit from pluripotency and alters Activin/Nodal, BMP, JNK-JUN, and WNT pathway signaling dynamics. These signaling changes influence downstream patterning of the gut tube, leading to improved pancreatic progenitor identity and decreased expression of markers associated with other endodermal lineages. Continued differentiation generates islets containing a higher percentage of β cells that exhibit improved maturation, insulin secretion, and ability to reverse hyperglycemia. Furthermore, this latA treatment reduces enterochromaffin cells in the final cell population and corrects differentiations from hPSC lines that otherwise fail to consistently produce pancreatic islets, highlighting the importance of cytoskeletal signaling at the onset of directed differentiation.

High-throughput discovery of inhibitory protein fragments with AlphaFold

Peptides can bind to specific sites on larger proteins and thereby function as inhibitors and regulatory elements. Peptide fragments of larger proteins are particularly attractive for achieving these functions due to their inherent potential to form native-like binding interactions. Recently developed experimental approaches allow for high-throughput measurement of protein fragment inhibitory activity in living cells. However, it has thus far not been possible to predict de novo which of the many possible protein fragments bind to protein targets, let alone act as inhibitors. We have developed a computational method, FragFold, that employs AlphaFold to predict protein fragment binding to full-length proteins in a high-throughput manner. Applying FragFold to thousands of fragments tiling across diverse proteins revealed peaks of predicted binding along each protein sequence. Comparisons with experimental measurements establish that our approach is a sensitive predictor of fragment function: Evaluating inhibitory fragments from known protein-protein interaction interfaces, we find 87% are predicted by FragFold to bind in a native-like mode. Across full protein sequences, 68% of FragFold-predicted binding peaks match experimentally measured inhibitory peaks. Deep mutational scanning experiments support the predicted binding modes and uncover superior inhibitory peptides in high throughput. Further, FragFold is able to predict previously unknown protein binding modes, explaining prior genetic and biochemical data. The success rate of FragFold demonstrates that this computational approach should be broadly applicable for discovering inhibitory protein fragments across proteomes.

Multi-lineage transcriptional and cell communication signatures define pathways in individuals at-risk for developing rheumatoid arthritis that initiate and perpetuate disease

Elevated anti-citrullinated protein antibodies (ACPA) levels in the peripheral blood are associated with an increased risk for developing rheumatoid arthritis (RA). Currently, no treatments are available that prevent progression to RA in these at-risk individuals. In addition, diverse pathogenic mechanisms underlying a common clinical phenotype in RA complicate therapy as no single agent is universally effective. We propose that a unifying set of transcription factor and their downstream pathways regulate a pro-inflammatory cell communication network, and that this network allows multiple cell types to serve as pathogenic drivers in at-risk individuals and in early RA. To test this hypothesis, we identified ACPA-positive at-risk individuals, patients with early ACPA-positive RA and matched controls. We measured single cell chromatin accessibility and transcriptomic profiles from their peripheral blood mononuclear cells. The datasets were then integrated to define key TF, as well as TF-regulated targets and pathways. A distinctive TF signature was enriched in early RA and at-risk individuals that involved key pathogenic mechanisms in RA, including SUMOylation, RUNX2, YAP1, NOTCH3, and β-Catenin Pathways. Interestingly, this signature was identified in multiple cell types, including T cells, B cells, and monocytes, and the pattern of cell type involvement varied among the at-risk and early RA participants, supporting our hypothesis. Similar patterns of individualized gene expression patterns and cell types were confirmed in single cell studies of RA synovium. Cell communication analysis revealed that the lineages displaying this RA TF signature deliver a common set of pro-inflammatory mediators to receiver cells that subsequently orchestrate rheumatoid inflammation. These cell-type-specific signature pathways could explain the personalized pathogenesis of RA and contribute to the diversity of clinical responses to targeted therapies. Furthermore, these data could provide opportunities for stratifying individuals at-risk for RA, and selecting therapies tailored for prevention or treatment of RA. Overall, this study supports a new paradigm to understand how a common clinical phenotype could arise from diverse pathogenic mechanisms and demonstrates the relevance of peripheral blood cells to synovial disease.

LINE1 elements at distal junctions of rDNA repeats regulate nucleolar organization in human embryonic stem cells

The nucleolus is a major subnuclear compartment where ribosomal DNA (rDNA) is transcribed and ribosomes are assembled. In addition, recent studies have shown that the nucleolus is a dynamic organizer of chromatin architecture that modulates developmental gene expression. rDNA gene units are assembled into arrays located in the p-arms of five human acrocentric chromosomes. Distal junctions (DJs) are ∼400 kb sequences adjacent to rDNA arrays that are thought to anchor them at the nucleolus, although the underlying regulatory elements remain unclear. Here we show that DJs display a dynamic chromosome conformation profile in human embryonic stem cells (hESCs). We identified a primate-specific, full-length insertion of the retrotransposon long interspersed nuclear element 1 (LINE1) in a conserved position across all human DJs. This DJ-LINE1 locus interacts with specific regions of the DJ and is upregulated in naïve hESCs. CRISPR-based deletion and interference approaches revealed that DJ-LINE1 contributes to nucleolar positioning of the DJs. Moreover, we found that the expression of DJ-LINE1 is required for maintenance of the structure and transcriptional output of the nucleolus in hESCs. Silencing of DJ-LINE1 leads to loss of self-renewal, disruption of the landscape of chromatin accessibility, and derepression of earlier developmental programs in naïve hESCs. This work uncovers specific LINE1 elements with a fundamental role in nucleolar organization in hESCs and provides new insights into how the nucleolus functions as a key genome-organizing hub.

Large-scale study of human memory for meaningful narratives

The statistical study of human memory requires large-scale experiments, involving many stimulus conditions and test subjects. While this approach has proven to be quite fruitful for meaningless material such as random lists of words, naturalistic stimuli, like narratives, have until now resisted such a large-scale study, due to the quantity of manual labor required to design and analyze such experiments. In this work, we develop a pipeline that uses large language models (LLMs) both to design naturalistic narrative stimuli for large-scale recall and recognition memory experiments, as well as to analyze the results. We performed online memory experiments with a large number of participants and collected recognition and recall data for narratives of different sizes. We found that both recall and recognition performance scale linearly with narrative length; however, for longer narratives, people tend to summarize the content rather than recalling precise details. To investigate the role of narrative comprehension in memory, we repeated these experiments using scrambled versions of the narratives. Although recall performance declined significantly, recognition remained largely unaffected. Recalls in this condition seem to follow the original narrative order rather than the actual scrambled presentation, pointing to a contextual reconstruction of the story in memory. Finally, using LLM text embeddings, we construct a simple measure for each clause based on semantic similarity to the whole narrative, that shows a strong correlation with recall probability. Overall, our work demonstrates the power of LLMs in accessing new regimes in the study of human memory, as well as suggesting novel psychologically informed benchmarks for LLM performance.

Exploring the impact of selenium-enriched peptides from yeast autolysate on dough properties: Insights into mechanisms from gluten perspectives

This study investigated the impact of Selenium (Se)-enriched yeast autolytic peptides (SeYAP) with different Se levels on dough properties as well as the related mechanism by focusing on gluten. SeYAP prolonged the dough's development time by up to 131 % and stability time by up to 28 %. It also decreased dough's viscoelasticity and rendered dough softer. Additionally, SeYAP diminished the binding capacity of dough to water and augmented the fluidity of water. Protein composition, disulfide bonds and fluorescence spectroscopy revealed that SeYAP could induce depolymerization of glutenin aggregate through sulfhydryl/disulfide bond exchange and hydrophobic interactions. Seven Se-enriched peptides were identified from the fraction with strong ability to depolymerize gluten. Specifically, six peptides contained selenocysteine, while another peptide contained selenomethionine. Molecular docking indicated that Se-enriched peptides could interact with amino acids (such as glutamine, tyrosine and proline) in gluten via hydrophobic interactions and/or hydrogen bonds.

AlphaFold and what is next: bridging functional, systems and structural biology

INTRODUCTION

The DeepMind's AlphaFold (AF) has revolutionized biomedical and biocience research by providing both experts and non-experts with an invaluable tool for predicting protein structures. However, while AF is highly effective for predicting structures of rigid and globular proteins, it is not able to fully capture the dynamics, conformational variability, and interactions of proteins with ligands and other biomacromolecules.

AREAS COVERED

In this review, we present a comprehensive overview of the latest advancements in 3D model predictions for biomacromolecules using AF. We also provide a detailed analysis its of strengths and limitations, and explore more recent iterations, modifications, and practical applications of this strategy. Moreover, we map the path forward for expanding the landscape of AF toward predicting structures of every protein and peptide, and their interactions in the proteome in the most physiologically relevant form. This discussion is based on an extensive literature search performed using PubMed and Google Scholar.

EXPERT OPINION

While significant progress has been made to enhance AF's modeling capabilities, we argue that a combined approach integrating both various in silico and in vitro methods will be most beneficial for the future of structural biology, bridging the gaps between static and dynamic features of proteins and their functions.

Artificial intelligence in peptide-based drug design

Protein-protein interactions (PPIs) are fundamental to a variety of biological processes, but targeting them with small molecules is challenging because of their large and complex interaction interfaces. However, peptides have emerged as highly promising modulators of PPIs, because they can bind to protein surfaces with high affinity and specificity. Nonetheless, computational peptide design remains difficult, hindered by the intrinsic flexibility of peptides and the substantial computational resources required. Recent advances in artificial intelligence (AI) are paving new paths for peptide-based drug design. In this review, we explore the advanced deep generative models for designing target-specific peptide binders, highlight key challenges, and offer insights into the future direction of this rapidly evolving field.

Binding mechanisms of intrinsically disordered proteins: Insights from experimental studies and structural predictions

Advances in the characterization of intrinsically disordered proteins (IDPs) have unveiled a remarkably complex and diverse interaction landscape, including coupled folding and binding, highly dynamic complexes, multivalent interactions, and even interactions between entirely disordered proteins. Here we review recent examples of IDP binding mechanisms elucidated by experimental techniques such as nuclear magnetic resonance spectroscopy, single-molecule Förster resonance energy transfer, and stopped-flow fluorescence. These techniques provide insights into the structural details of transition pathways and complex intermediates, and they capture the dynamics of IDPs within complexes. Furthermore, we discuss the growing role of artificial intelligence, exemplified by AlphaFold, in identifying interaction sites within IDPs and predicting their bound-state structures. Our review highlights the powerful complementarity between experimental methods and artificial intelligence-based approaches in advancing our understanding of the intricate interaction landscape of IDPs.

De novo design of peptide binders to conformationally diverse targets with contrastive language modeling

Designing binders to target undruggable proteins presents a formidable challenge in drug discovery. In this work, we provide an algorithmic framework to design short, target-binding linear peptides, requiring only the amino acid sequence of the target protein. To do this, we propose a process to generate naturalistic peptide candidates through Gaussian perturbation of the peptidic latent space of the ESM-2 protein language model and subsequently screen these novel sequences for target-selective interaction activity via a contrastive language-image pretraining (CLIP)-based contrastive learning architecture. By integrating these generative and discriminative steps, we create a Peptide Prioritization via CLIP (PepPrCLIP) pipeline and validate highly ranked, target-specific peptides experimentally, both as inhibitory peptides and as fusions to E3 ubiquitin ligase domains. PepPrCLIP-derived constructs demonstrate functionally potent binding and degradation of conformationally diverse, disease-driving targets in vitro. In total, PepPrCLIP empowers the modulation of previously inaccessible proteins without reliance on stable and ordered tertiary structures.

HPV integration in head and neck cancer: downstream splicing events and expression ratios linked with poor outcomes

HPV integration (HPVint) is associated with carcinogenesis and tumor progression in HPV-associated cancers, including head and neck squamous cell carcinomas (HNSCC). While its impact on human DNA has been well characterized, its relationship with clinical outcomes remains unconfirmed. Here we investigate the consequences of HPVint both with respect to human and HPV characteristics by analyzing 261 HPV-associated HNSCC bulk and single-cell RNA-seq samples from five cohorts, and DNA HPVint events from 102 HPV+ participants in two of the cohorts. By leveraging this large meta-cohort, we first reveal an oncogenic network based on the recurrent HPV integration locations in HNSCC. We then classify HPVint-positive (HPVint(+)) participants by HPV RNA features, specifically based on spliced HPV-human fusion transcripts and ratios of HPV gene transcripts, showing that subsets of participants have worse clinical outcomes. Our analyses, focused mainly on RNA instead of DNA, expand our understanding of the carcinogenic mechanisms of HPVint, partially addressing the conflicting findings of whether HPVint is associated with aggressive phenotypes and worse clinical consequences, and provide potential biomarkers to advance precision oncology in HPV-associated HNSCC.

Hybridization Design and High-Throughput Screening of Peptides with Immunomodulatory and Antioxidant Activities

With the increasing recognition of the role of immunomodulation and oxidative stress in various diseases, designing peptides with both immunomodulatory and antioxidant activities has emerged as a promising therapeutic strategy. In this study, a hybridization design was applied as a powerful method to obtain multifunctional peptides. A total of 40 peptides with potential immunomodulatory and antioxidant activities were designed and screened. First, molecular docking was employed to screen peptides with a high binding affinity to MD2, a key receptor protein in the NFκB immune pathway. For the in vitro high-throughput screening, we constructed a reporter gene-based stable cell line, IPEC-J2-Lucia ARE cells, which was subsequently used to screen peptides with antioxidant activity. Furthermore, the biocompatibility, immunomodulatory, and antioxidant activities of these peptides were assessed. Among the candidates, the hybrid peptide VA exhibited the strongest immune-enhancing activity through the activation of the NF-κB pathway and significant antioxidant activity via the Nrf2-ARE pathway. Additionally, VA demonstrated protective effects against H2O2-induced oxidative damage in HepG2 cells. This study not only demonstrates the potential of peptide hybridization, but also develops a screening platform for multifunctional peptides. It provides a new tool for the treatment of autoimmune diseases and oxidative stress-related diseases.

Transcription Factor-Wide Association Studies to Identify Functional SNPs in Alzheimer's Disease

Alzheimer's disease (AD) is a progressive neurodegenerative disorder with profound global impact. While genome-wide association studies (GWAS) have revealed genomic variants linked to AD, their translational impact has been limited due to challenges in interpreting the identified genetic associations. To address this challenge, we have devised a novel approach termed transcription factor-wide association studies (TF-WAS). By integrating the GWAS, expression quantitative trait loci, and transcriptome analyses, we selected 30 AD single nucleotide polymorphisms (SNPs) in noncoding regions that are likely to be functional. Using human transcription factor (TF) microarrays, we have identified 90 allele-specific TF interactions with 53 unique TFs. We then focused on several interactions involving SMAD4 and further validated them using electrophoretic mobility shift assay, luciferase, and chromatin immunoprecipitation on engineered genetic backgrounds (female cells). This approach holds promise for unraveling the intricacies of not just AD, but any complex disease with available GWAS data, providing insight into underlying molecular mechanisms and clues toward potential therapeutic targets.

Systemic Metabolic Alterations after Aneurysmal Subarachnoid Hemorrhage: A Plasma Metabolomics Approach

Background

Aneurysmal subarachnoid hemorrhage (aSAH) causes systemic changes that contribute to delayed cerebral ischemia (DCI) and morbidity. Circulating metabolites reflecting underlying pathophysiological mechanisms warrant investigation as biomarker candidates.

Methods

Blood samples, prospectively collected within 24 hours (T1) of admission and 7-days (T2) post ictus, from patients with acute aSAH from two tertiary care centers were retrospectively analyzed. Samples from healthy subjects and patients with non-neurologic critical illness served as controls. A validated external analysis platform was used to perform untargeted metabolomics. Bioinformatics analyses were conducted to identify metabolomic profiles defining each group and delineate metabolic pathways altered in each group. Machine learning (ML) models were developed incorporating key metabolites to improve DCI prediction.

Results

Among 70 aSAH, 30 healthy control, and 17 sick control subjects, a total of 1,117 metabolites were detected. Groups were matched among key clinical variables. DCI occurred in 36% of aSAH subjects, and poor functional outcome was observed in 70% at discharge. Metabolomic profiles readily discriminated the groups. aSAH subjects demonstrated a robust mobilization of lipid metabolites, with increased levels of free fatty acids (FFAs), mono- and diacylglycerols (MAG, DAG) compared with both control groups. aSAH subjects also had decreased circulating amino acid derived metabolites, consistent with increased catabolism. DCI was associated with increased sphingolipids (sphingosine and sphinganine) and decreased acylcarnitines and S-adenosylhomocysteine at T1. Decreased lysophospholipids and acylcarnitines were associated with poor outcomes. Incorporating metabolites into ML models improved prediction of DCI compared with clinical variables alone.

Conclusions

Profound metabolic shifts occur after aSAH with characteristic increases in lipid and decreases in amino acid metabolites. Key lipid metabolites associated with outcomes (sphingolipids, lysophospholipids, and acylcarnitines) provide insight into systemic changes driving secondary complications. These metabolites may also prove to be useful biomarkers to improve prognostication and personalize aSAH care.

Recombination, truncation and horizontal transfer shape the diversity of cytoplasmic incompatibility patterns

Wolbachia are endosymbiotic bacteria inducing various reproductive manipulations of which cytoplasmic incompatibility (CI) is the most common. CI leads to reduced embryo viability in crosses between males carrying Wolbachia and uninfected females or those carrying an incompatible symbiont strain. In the mosquito Culex pipiens, the Wolbachia wPip causes highly complex crossing patterns. This complexity is linked to the amplification and diversification of the CI causal genes, cidA and cidB, with polymorphism located in the CidA-CidB interaction regions. We previously showed correlations between the identity of gene variants and CI patterns. However, these correlations were limited to specific crosses, and it is still unknown whether cid gene polymorphism in males' and females' Wolbachia can explain and predict the wide range of crossing types observed in C. pipiens. Taking advantage of a new method enabling full-gene acquisition, we sequenced complete cid repertoires from 45 wPip strains collected worldwide. We demonstrated that the extensive diversity of cid genes arises from recombination and horizontal transfers. We uncovered further cidB polymorphism outside the interface regions and strongly correlated with CI patterns. Most importantly, we showed that in every wPip genome, all but one cidB variant are truncated. Truncated cidBs located in palindromes are partially or completely deprived of their deubiquitinase domain, crucial for CI. The identity of the sole full-length cidB variant seems to dictate CI patterns, irrespective of the truncated cidBs present. Truncated CidBs exhibit reduced toxicity and stability in Drosophila cells, which potentially hinders their loading into sperm, essential for CI induction.

Engineering of CRISPR-Cas PAM recognition using deep learning of vast evolutionary data

CRISPR-Cas enzymes must recognize a protospacer-adjacent motif (PAM) to edit a genomic site, significantly limiting the range of targetable sequences in a genome. Machine learning-based protein engineering provides a powerful solution to efficiently generate Cas protein variants tailored to recognize specific PAMs. Here, we present Protein2PAM, an evolution-informed deep learning model trained on a dataset of over 45,000 CRISPR-Cas PAMs. Protein2PAM rapidly and accurately predicts PAM specificity directly from Cas proteins across Type I, II, and V CRISPR-Cas systems. Using in silico deep mutational scanning, we demonstrate that the model can identify residues critical for PAM recognition in Cas9 without utilizing structural information. As a proof of concept for protein engineering, we employ Protein2PAM to computationally evolve Nme1Cas9, generating variants with broadened PAM recognition and up to a 50-fold increase in PAM cleavage rates compared to the wild-type under in vitro conditions. This work represents the first successful application of machine learning to achieve customization of Cas enzymes for alternate PAM recognition, paving the way for personalized genome editing.

Accurate sequence-to-affinity models for SH2 domains from multi-round peptide binding assays coupled with free-energy regression

Short linear peptide motifs play important roles in cell signaling. They can act as modification sites for enzymes and as recognition sites for peptide binding domains. SH2 domains bind specifically to tyrosine-phosphorylated proteins, with the affinity of the interaction depending strongly on the flanking sequence. Quantifying this sequence specificity is critical for deciphering phosphotyrosine-dependent signaling networks. In recent years, protein display technologies and deep sequencing have allowed researchers to profile SH2 domain binding across thousands of candidate ligands. Here, we present a concerted experimental and computational strategy that improves the predictive power of SH2 specificity profiling. Through multi-round affinity selection and deep sequencing with large randomized phosphopeptide libraries, we produce suitable data to train an additive binding free energy model that covers the full theoretical ligand sequence space. Our models can be used to predict signaling network connectivity and the impact of missense variants in phosphoproteins on SH2 binding.

Reliable protein-protein docking with AlphaFold, Rosetta, and replica-exchange

Despite the recent breakthrough of AlphaFold (AF) in the field of protein sequence-to-structure prediction, modeling protein interfaces and predicting protein complex structures remains challenging, especially when there is a significant conformational change in one or both binding partners. Prior studies have demonstrated that AF-multimer (AFm) can predict accurate protein complexes in only up to 43% of cases.1 In this work, we combine AlphaFold as a structural template generator with a physics-based replica exchange docking algorithm to better sample conformational changes. Using a curated collection of 254 available protein targets with both unbound and bound structures, we first demonstrate that AlphaFold confidence measures (pLDDT) can be repurposed for estimating protein flexibility and docking accuracy for multimers. We incorporate these metrics within our ReplicaDock 2.0 protocol2to complete a robust in-silico pipeline for accurate protein complex structure prediction. AlphaRED (AlphaFold-initiated Replica Exchange Docking) successfully docks failed AF predictions including 97 failure cases in Docking Benchmark Set 5.5. AlphaRED generates CAPRI acceptable-quality or better predictions for 63% of benchmark targets. Further, on a subset of antigen-antibody targets, which is challenging for AFm (20% success rate), AlphaRED demonstrates a success rate of 43%. This new strategy demonstrates the success possible by integrating deep-learning based architectures trained on evolutionary information with physics-based enhanced sampling. The pipeline is available at github.com/Graylab/AlphaRED.

Artificial intelligence in drug development

Drug development is a complex and time-consuming endeavor that traditionally relies on the experience of drug developers and trial-and-error experimentation. The advent of artificial intelligence (AI) technologies, particularly emerging large language models and generative AI, is poised to redefine this paradigm. The integration of AI-driven methodologies into the drug development pipeline has already heralded subtle yet meaningful enhancements in both the efficiency and effectiveness of this process. Here we present an overview of recent advancements in AI applications across the entire drug development workflow, encompassing the identification of disease targets, drug discovery, preclinical and clinical studies, and post-market surveillance. Lastly, we critically examine the prevailing challenges to highlight promising future research directions in AI-augmented drug development.

PAbFold: Linear Antibody Epitope Prediction using AlphaFold2

Defining the binding epitopes of antibodies is essential for understanding how they bind to their antigens and perform their molecular functions. However, while determining linear epitopes of monoclonal antibodies can be accomplished utilizing well-established empirical procedures, these approaches are generally labor- and time-intensive and costly. To take advantage of the recent advances in protein structure prediction algorithms available to the scientific community, we developed a calculation pipeline based on the localColabFold implementation of AlphaFold2 that can predict linear antibody epitopes by predicting the structure of the complex between antibody heavy and light chains and target peptide sequences derived from antigens. We found that this AlphaFold2 pipeline, which we call PAbFold, was able to accurately flag known epitope sequences for several well-known antibody targets (HA / Myc) when the target sequence was broken into small overlapping linear peptides and antibody complementarity determining regions (CDRs) were grafted onto several different antibody framework regions in the single-chain antibody fragment (scFv) format. To determine if this pipeline was able to identify the epitope of a novel antibody with no structural information publicly available, we determined the epitope of a novel anti-SARS-CoV-2 nucleocapsid targeted antibody using our method and then experimentally validated our computational results using peptide competition ELISA assays. These results indicate that the AlphaFold2-based PAbFold pipeline we developed is capable of accurately identifying linear antibody epitopes in a short time using just antibody and target protein sequences. This emergent capability of the method is sensitive to methodological details such as peptide length, AlphaFold2 neural network versions, and multiple-sequence alignment database. PAbFold is available at https://github.com/jbderoo/PAbFold.

An endogenous cluster of target-directed microRNA degradation sites induces decay of distinct microRNA families

While much is known about miRNA biogenesis and canonical miRNA targeting, relatively less is understood about miRNA decay. The major miRNA decay pathway in metazoans is mediated through target-directed miRNA degradation (TDMD), in which certain RNAs can "trigger" miRNA decay. All known triggers for TDMD base pair with the miRNA seed, and extensively base pair on the miRNA 3' end, a pattern that supposedly induces a TDMD-competent conformational change of Argonaute (Ago), allowing for miRNA turnover. Here, we utilized Ago1-CLASH to find that the Drosophila transcript Kah contains at least two triggers, a "trigger cluster", against miR-9b and the miR-279 family. One of these triggers contains minimal/non-canonical 3' end base pairing but is still sufficient to induce TDMD of the entire miR-279 family. We found that these clustered triggers likely lack cooperativity, the minimal 3' pairing is required for miR-279 family turnover, and probed the in-cell RNA structure of the Kah trigger cluster. Overall, this study expands the list of endogenous triggers and the unexpectedly complex regulatory network governing miRNA degradation.

Human genetics implicate thromboembolism in the pathogenesis of long COVID in individuals of European ancestry

SARS-CoV-2 infection can result in long COVID, characterized by post-acute symptoms from multiple organs. Current hypotheses on mechanisms underlying long COVID include persistent inflammation and thromboembolism; however, compelling evidence from humans is limited and causal associations remain unclear. Here, we tested the association of thromboembolism-related genetic variants with long COVID in the Long COVID Host Genetics Initiative ( n cases =3,018; n controls =994,582). Primary analyses revealed that each unit increase in the log-odds of genetically predicted venous thromboembolism risk was associated with 1.21-fold odds of long COVID (95%CI: 1.08-1.35; P =1.2 × 10 -3 ). This association was independent of acute COVID-19 severity, robust across genetic instruments and methods, and replicated in external datasets for both venous thromboembolism and long COVID. Downstream analyses using gene-specific instruments, along with protein and gene expression data, suggested the protease-activated receptor 1 (PAR-1) as a potential molecular contributor to long COVID. These findings provide human genetic evidence implicating thromboembolism in long COVID pathogenesis. .

Massively parallel genetic perturbation reveals the energetic architecture of an amyloid beta nucleation reaction

Amyloid protein aggregates are pathological hallmarks of more than fifty human diseases but how soluble proteins nucleate to form amyloids is poorly understood. Here we use combinatorial mutagenesis, a kinetic selection assay, and machine learning to massively perturb the energetics of the nucleation reaction of amyloid beta (Aβ42), the protein that aggregates in Alzheimer's disease. In total, we quantify the nucleation rates of >140,000 variants of Aβ42. This allows us to accurately quantify the changes in reaction activation energy for all possible amino acid substitutions in a protein for the first time and, in addition, to quantify >600 energetic interactions between mutations. The data reveal the simple and interpretable genetic architecture of an amyloid nucleation reaction. Strikingly, strong energetic couplings are rare and identify a subset of structural contacts in mature fibrils. Together with the activation energy changes, this strongly suggests that the Aβ42 nucleation reaction transition state is structured in a short C-terminal region, providing a structural model for the reaction that may initiate Alzheimer's disease. We believe this approach can be widely applied to probe the energetics and transition state structures of protein reactions.

Plasticity of Human Microglia and Brain Perivascular Macrophages in Aging and Alzheimer's Disease

The complex roles of myeloid cells, including microglia and perivascular macrophages, are central to the neurobiology of Alzheimer's disease (AD), yet they remain incompletely understood. Here, we profiled 832,505 human myeloid cells from the prefrontal cortex of 1,607 unique donors covering the human lifespan and varying degrees of AD neuropathology. We delineated 13 transcriptionally distinct myeloid subtypes organized into 6 subclasses and identified AD-associated adaptive changes in myeloid cells over aging and disease progression. The GPNMB subtype, linked to phagocytosis, increased significantly with AD burden and correlated with polygenic AD risk scores. By organizing AD-risk genes into a regulatory hierarchy, we identified and validated MITF as an upstream transcriptional activator of GPNMB, critical for maintaining phagocytosis. Through cell-to-cell interaction networks, we prioritized APOE-SORL1 and APOE-TREM2 ligand-receptor pairs, associated with AD progression. In both human and mouse models, TREM2 deficiency disrupted GPNMB expansion and reduced phagocytic function, suggesting that GPNMB's role in neuroprotection was TREM2-dependent. Our findings clarify myeloid subtypes implicated in aging and AD, advancing the mechanistic understanding of their role in AD and aiding therapeutic discovery.

Polyfunctional IL-21+ IFNG+ T follicular helper cells contribute to checkpoint inhibitor diabetes mellitus and can be targeted by JAK inhibitor therapy

Immune checkpoint inhibitors (ICI) have revolutionized cancer therapy, but their use is limited by the development of autoimmunity in healthy tissues as a side effect of treatment. Such immune-related adverse events (IrAE) contribute to hospitalizations, cancer treatment interruption and even premature death. ICI-induced autoimmune diabetes mellitus (ICI-T1DM) is a life-threatening IrAE that presents with rapid pancreatic beta-islet cell destruction leading to hyperglycemia and life-long insulin dependence. While prior reports have focused on CD8+ T cells, the role for CD4+ T cells in ICI-T1DM is less understood. Here, we identify expansion CD4+ T follicular helper (Tfh) cells expressing interleukin 21 (IL-21) and interferon gamma (IFNG) as a hallmark of ICI-T1DM. Furthermore, we show that both IL-21 and IFNG are critical cytokines for autoimmune attack in ICI-T1DM. Because IL-21 and IFNG both signal through JAK-STAT pathways, we reasoned that JAK inhibitors (JAKi) may protect against ICI-T1DM. Indeed, JAKi provide robust in vivo protection against ICI-T1DM in a mouse model that is associated with decreased islet-infiltrating Tfh cells. Moreover, JAKi therapy impaired Tfh cell differentiation in patients with ICI-T1DM. These studies highlight CD4+ Tfh cells as underrecognized but critical mediators of ICI-T1DM that may be targeted with JAKi to prevent this grave IrAE.

Wideband Tuning and Deep-Tissue Spectral Detection of Indium Phosphide Nano-Laser Particles

Laser particles (LPs) emitting narrowband spectra across wide spectral ranges are highly promising for high-multiplex optical barcoding. Here, we present LPs based on indium phosphide (InP) nanodisks, operating in the near-infrared wavelength range of 740-970 nm. Utilizing low-order whispering gallery resonance modes in size-tuned nanodisks, we achieved an ultrawide color palette with 27% bandwidth utilization and nanometer-scale linewidth. The minimum laser size was 430 nm in air and 560 nm within the cytoplasm, operating at mode order 4 or 5. We further demonstrated spectral detection of laser peaks with high signal-to-background ratios in highly-scattering media, including 1-cm-thick chicken breast tissue and blood vessels in live mice.

ESM-scan-A tool to guide amino acid substitutions

Protein structure prediction and (re)design have gone through a revolution in the last 3 years. The tremendous progress in these fields has been almost exclusively driven by readily available machine learning algorithms applied to protein folding and sequence design problems. Despite these advancements, predicting site-specific mutational effects on protein stability and function remains an unsolved problem. This is a persistent challenge, mainly because the free energy of large systems is very difficult to compute with absolute accuracy and subtle changes to protein structures are hard to capture with computational models. Here, we describe the implementation and use of ESM-Scan, which uses the ESM zero-shot predictor to scan entire protein sequences for preferential amino acid changes, thus enabling in silico deep mutational scanning experiments. We benchmark ESM-Scan on its predictive capabilities for stability and functionality of sequence changes using three publicly available datasets and proceed by experimentally testing the tool's performance on a challenging test case of a blue-light-activated diguanylate cyclase from Methylotenera species (MsLadC), where it accurately predicted the importance of a highly conserved residue in a region involved in allosteric product inhibition. Our experimental results show that the ESM-zero shot model is capable of inferring the effects of a set of amino acid substitutions in their correlation between predicted fitness and experimental results. ESM-Scan is publicly available at https://huggingface.co/spaces/thaidaev/zsp.

In silico screening of protein-binding peptides with an application to developing peptide inhibitors against antibiotic resistance

The field of therapeutic peptides is experiencing a surge, fueled by their advantageous features. These include predictable metabolism, enhanced safety profile, high selectivity, and reduced off-target effects compared with small-molecule drugs. Despite progress in addressing limitations associated with peptide drugs, a significant bottleneck remains: the absence of a large-scale in silico screening method for a given protein target structure. Such methods have proven invaluable in accelerating small-molecule drug discovery. The high flexibility of peptide structures and the large diversity of peptide sequences greatly hinder the development of urgently needed computational methods. Here, we report a method called MDockPeP2_VS to address these challenges. It integrates molecular docking with structural conservation between protein folding and protein-peptide binding. Briefly, we discovered that when the interfacial residues are conserved, a sequence fragment derived from a monomeric protein exhibits a high propensity to bind a target protein with a similar conformation. This valuable insight significantly reduces the search space for peptide conformations, resulting in a substantial reduction in computational time and making in silico peptide screening practical. We applied MDockPeP2_VS to develop peptide inhibitors targeting the TEM-1 β-lactamase of Escherichia coli, a key mechanism behind antibiotic resistance in gram-negative bacteria. Among the top 10 peptides selected from in silico screening, TF7 (KTYLAQAAATG) showed significant inhibition of β-lactamase activity with a K i value of 1.37 ± 0.37 µM. This fully automated, large-scale structure-based in silico peptide screening software is available for free download at https://zougrouptoolkit.missouri.edu/mdockpep2_vs/download.html.