共找到 20 条结果
Macromolecular assemblies underpin essential cellular processes, yet their structural characterization remains challenging. Integrative modeling provides an approach for determining structures of macromolecular assemblies, combining diverse experimental data with physical principles, the statistics of previous structures, and prior models. There is a growing interest in leveraging the implicit structural knowledge learned by artificial intelligence-based structure-prediction methods such as AlphaFold (AF), for integrative modeling. Here, we discuss recent methods that combine AF with experimental data for integrative modeling in four ways: validating AF-based ensembles with experimental data; combining structural priors from AF with experimental data; fine-tuning AF with experimental data; and incorporating experimental data at inference time. We also outline key challenges in integrative structure determination using AF.
Single-molecule force spectroscopy (SMFS) techniques initially emerged as a new method to probe protein biophysics, often providing complementary insights to biochemical bulk experiments. Over time, however, advances in instrumentation and the growing recognition that mechanical forces are integral to biological function have progressively redirected its use toward exploring protein systems operating in mechanically active environments. In this review, we highlight recent applications of SMFS that shed light on how force regulates protein function, spanning diverse biological systems like cotranslational folding, protein degradation, and cellular adhesion proteins. Beyond allowing us to manipulate individual molecules, SMFS uniquely recreates the mechanical conditions under which many proteins operate, revealing mechanistic details inaccessible to traditional protein characterization methods. Looking ahead, ongoing innovation, both in instrumentation and in the integration of SMFS with complementary techniques, is bringing the field closer to mimicking physiologically relevant conditions. These developments are opening new avenues for recognizing mechanical force as a central regulator of biological systems.
During eukaryotic translation initiation, initiation factor proteins and the ribosomal small subunit undergo binding and dissociation reactions and conformational rearrangements that properly assemble a ribosome at the start codon of a messenger RNA. Building on extensive genetic and biochemical studies, single-molecule fluorescence experiments are revealing the time-dependent pathways of factor binding to, and dissociation from, the ribosomal small subunit and messenger RNA during initiation. Nonetheless, essential binding and/or dissociation events, conformational rearrangements, and the coupling between binding and conformational changes remain kinetically uncharacterized. Here, we summarize the status of single-molecule investigations of initiation and advocate for integrating single-molecule microscopy, structural biology, and molecular simulations to enable a time-dependent, molecular description of this fundamental step in gene expression.
A small number of residues at protein-protein interfaces, commonly referred to as hotspots, dominate binding free energy and play a decisive role in stabilizing protein complexes. Identifying these residues is central to understanding the energetic architecture of protein-protein interactions and to developing strategies for therapeutic intervention. Although experimental approaches such as alanine scanning have provided critical insights, they are often impractical for large or dynamic systems. This has positioned computational approaches at the forefront of hotspot analysis. This review highlights recent developments in molecular dynamics simulations and machine-learning-based predictors for hotspot identification, discusses current challenges, and outlines emerging directions in the field. Finally, we suggest that combining these complementary approaches could offer a powerful framework for capturing the dynamic and energetic complexity of protein interfaces, making hotspot predictions more robust and interpretable.
Understanding how proteins function requires a description of their structural dynamics. Accessing these structural dynamics across time and length scales calls for different methods and often their combination. Here we review a combination of two highly synergistic and complementary methods-single-molecule Förster resonance energy transfer (smFRET) and hydrogen-deuterium exchange mass spectrometry (HDX-MS). We describe how they complement each other in the ability to resolve structural and dynamical heterogeneity, coverage of protein residues, and types of protein motions they probe. Together, they access a wide range of protein dynamics from global rigid body motions to local unfolding. This synergy has led to the elucidation of activation and regulatory mechanisms in signaling and transcription, which we review in five case studies, highlighting the key role these two methods play in advancing dynamic structural biology.
The conformational ensemble of a protein and its corresponding probabilities and dynamics are crucial determinants of its function, but are difficult to access with traditional experimental and computational technologies. This review examines the landscape of machine learning for modeling protein conformational ensembles. We categorize computational methods into three classes: AlphaFold-based approaches that modify the input multiple sequence alignment, score-based generative models that use diffusion or flow-matching algorithms, and protein language models that link sequence evolution with sequential dynamics. We discuss the data available for training and benchmarking, including molecular dynamics simulations and experimental repositories. We highlight current limitations in the field, including the lack of standardization in benchmarking and the high variability of mechanisms and environmental conditions that challenge current methods. Drawing lessons from the success of AlphaFold, we identify key opportunities for further improvement, including accurate modeling of kinetics and thermodynamics, and linking model uncertainty with targeted collection of new data.
Cryogenic electron tomography (cryoET) offers unparalleled views into the molecular architecture of cells. As no stains or fixation are used, electrons scatter off the native atoms, and all molecules contribute to the final tomogram. As a result, it can be challenging to identify proteins of interest, especially inside a crowded cellular environment. Recent developments in molecular tags for cryoET provide several options for identifying proteins in reconstructed tomograms, but these are often not appropriate for finding an area of interest when collecting data. To increase the utility and throughput of cryoET, future approaches should combine correlative light and electron microscopy (CLEM) with tagging, so that a single modification can be used at small and large spatial scales. Automation of the detection of tags in tomograms and correlation between imaging modalities using machine learning methods will help increase the throughput of these methods, making them more suitable for rare events or structure determination by sub-tomogram averaging.
G-protein-coupled receptors (GPCRs) are a large family of membrane proteins that mediate cellular responses to diverse stimuli and serve as targets for ∼35 % of Food and Drug Administration-approved drugs. Their structural complexity, conformational heterogeneity, and membrane embedding have historically hindered experimental characterization, although advances in crystallization and cryogenic electron microscopy have expanded access to high-resolution receptor structures. In parallel, artificial intelligence (AI) has transformed protein modeling and drug discovery as recognized by the 2024 Nobel Prize in Chemistry. This minireview highlights recent applications of AI to GPCR research (2023-2025), including structure prediction, virtual screening, generative design of small molecules and protein binders, mechanistic studies using molecular dynamics, and systems-level analyses. Together, these approaches are reshaping GPCR biology and accelerating next-generation drug discovery.
Water-soluble polymers are widely used as model crowders, yet their effects on proteins are often interpreted using frameworks developed for rigid spherical depletants. Here we review polymer crowding from the perspective of scaling theory, emphasizing how polymer-specific length scales govern protein-polymer interactions across concentration regimes. In dilute solutions, depletion is set by the polymer radius of gyration and scales linearly with concentration. Above the overlap concentration, c∗, the relevant length becomes the correlation length, ξ(c), which defines the mesh size and controls both the magnitude and range of interactions. Protein association, folding, and intrinsically disordered protein structure follow distinct scaling regimes determined by the ratio of protein size to ξ. Deviations from classical predictions arise from polymer connectivity and soft protein-polymer interactions. The polymer-scaling perspective provides a unified framework linking polymer physics to protein thermodynamics in crowded environments.
Generative artificial intelligence models learn probability distributions from data and produce novel samples that capture the salient properties of their training sets. Proteins are particularly attractive for such approaches given their abundant data and the versatility of their representations, ranging from sequences to structures and functions. This versatility has motivated the rapid development of generative models for protein design, enabling the generation of functional proteins and enzymes with unprecedented success. However, because these models mirror their training distribution, they tend to sample from its most probable modes, while low-probability regions, often encoding valuable properties, remain underexplored. To address this challenge, recent work has proposed strategies for steering generative models toward user-specified properties. In this review, we survey and categorize these strategies, distinguishing approaches that modify model parameters, such as reinforcement learning or supervised fine-tuning, from those that keep the model's parameters fixed, including conditional generation, retrieval-augmented strategies, Bayesian guidance, and tailored sampling methods. Together, these developments are beginning to enable the steering of generative models toward proteins with desired properties.
Nucleocytoplasmic transport relies on targeting signals within cargo polypeptides, typically as short linear motifs but sometimes as folded domains. These signals are recognized by the Karyopherin-β (Kap) family of importins, exportins, and biportins. Despite the number of Kaps, only a few linear signal classes are well-defined: the classical nuclear localization signal (cNLS) recognized by importin-α (IMPα), which in turn binds IMPβ to form the IMPα/β heterodimer, the IMPβ-binding domain, the Pro-Tyr NLS of transportin-1 (TNPO1/Kapβ2), the IK-NLS of Kap121/importin-5, and the RS/E- and RSY-NLSs of TNPO3, along with the classical nuclear export signal (NES) of exportin-1 (XPO1/CRM1) and the phosphorylated NES of yeast Msn5. This review summarizes recent structural and biochemical advances that define these signals and their recognition rules and highlights the remaining gaps in our understanding of linear signals across the Kap family.
Single-molecule experiments have become an integral part of modern structural biology. Unlike other methods, single-molecule Förster resonance energy transfer (smFRET) spectroscopy opens direct access to distance-based temporal trajectories of protein motions. Recent innovations in analysing smFRET experiments with correlation and photon-trajectory based methods have pushed the time resolution of dynamics to much faster than milliseconds. Here, we review these methods, together with their most recent applications and their impact on our understanding of the function of proteins. Important current topics range from the dynamics of intrinsically disordered proteins in complex with their binding partners or in biomolecular condensates, to the conformational dynamics of proteins during their function, from enzymes to molecular machines. We focus particularly on the determination of the timescales of motions and how the utmost information can be gleaned from single-molecule data at the single-photon level.
Intrinsically disordered proteins (IDPs) populate heterogeneous conformational ensembles, making them particularly sensitive to the crowded intracellular environment. Defining how molecular crowding reshapes these ensembles is therefore essential for bridging in vitro biophysical observations with cellular function. Recent studies have shown that crowding does not simply drive non-specific compaction. Instead, it remodels IDP conformational ensembles through competing entropic, enthalpic and solvent-mediated contributions, giving rise to diverse and sequence-dependent outcomes. In this review, we summarize recent experimental and computational advances that reveal how distinct classes of crowders modulate IDP conformations and how these effects are further tuned in cellular environments. We also discuss the consequences of crowding-induced ensemble remodeling for molecular recognition, biomolecular phase separation, and aggregation. Together, these findings establish molecular crowding as a key determinant of IDP conformational landscapes and functional behavior in complex biological settings.
Single-molecule microscopy has transformed our view of biomolecular condensates-membraneless organelles that organize cellular biochemistry and are frequently dysregulated in disease-revealing them not as simple liquid droplets, but as spatially heterogeneous and percolated networks that can undergo time-dependent physical aging and gelation. Here, we summarize how single-particle tracking, single-molecule-fluorescence resonance energy transfer and super-resolution microscopy resolve molecular motion, confinement, and conformational dynamics to link nanoscale behaviors to mesoscale condensate material properties and biological function. In vitro reconstitution affords mechanistic control, whereas emerging live-cell imaging probes physiological context. Photobleaching, phototoxicity, and autofluorescence remain challenges that are increasingly mitigated by optimized fluorophore and label-free approaches. Concurrently, deep-learning pipelines automate analysis and expose hidden heterogeneities. Further integrating artificial intelligence and imaging advances will be essential for decoding condensate structure-function relationships in health and disease.
Glycosylation can be critical for determining the structure and functions of proteins, but it is often neglected, leading to significant knowledge gaps in our understanding of biology. The inherent heterogeneity of glycans presents technical challenges to glycoprotein characterisation and impedes the representation of intact glycoprotein structures. Here we discuss how glycan heterogeneity constitutes a fundamental property of glycoproteins and acts as a remarkably powerful strategy for modulating biological function on the fly, complementing the rigidity of the genome. We present recent examples of how integrating mass spectrometry glycoproteomics and glycomics, structural biology, and molecular dynamics simulation data can bring glycans into a 3D structural framework, providing a unique perspective into the roles of glycosylation and potentially accelerating the design of glycoprotein biologics. Strengths and limitations of these approaches are also highlighted.
Protein function emerges from ensembles of interconverting conformations, presenting challenges beyond static structure prediction. Although recent advances in generative artificial intelligence (AI) have transformed native-fold prediction, capturing conformational landscapes and binding-associated structural transitions remains a central limitation for mechanistic biology and structure-based drug discovery. Emerging generative models now aim to learn dynamic conformational distributions directly, enabling ensemble generation, ligand-responsive receptor sampling, and pathway-level inference. However, predictive fidelity is still constrained by limited physical grounding, incomplete kinetic realism, data imbalance, and uncertain calibration. This review summarizes key developments from 2023 to 2025, examines their methodological and practical limitations, and discusses how generative AI may evolve into a reliable framework for dynamic structural biology and mechanism-guided drug design.
Cryo-electron microscopy (cryo-EM) is transitioning from determining structures of isolated proteins in vitro to visualizing macromolecular architecture directly in situ. Conventional in situ approaches, primarily relying on cryo-electron tomography combined with subtomogram averaging, are often limited in resolution due to complex workflows, cumulative errors in processing, and low data throughput. Emerging in situ single-particle cryo-EM methods address these limitations by collecting high-dose, untilted images of cellular lamellae. Using high-resolution templates for particle identification and refinement, these methods have significantly advanced both data throughput and achievable resolution. This review systematically outlines the principles, workflow, and advantages of in situ single-particle methods, highlights their key applications, and discusses future perspectives.
Understanding how tumor cells navigate crowded spatial environments to drive progression and how to deter them is challenging. Intracellularly, dynamic protein ensembles link genotype to phenotype. Dysregulation of these ensembles-driven by overexpression and mutational variants-alters the conformational landscapes, shifts cell states, and reshapes cell fate decisions. This diversity, spanning from the molecular level to the tumor microenvironment, triggers resistance mechanisms precipitating efforts to engineer effective combination strategies. Here, we underscore these transient cell states in migration and tissue adaptation, which depend on transcriptomic and signaling compatibility. Our spatial biology outlook envisions a map for designing drug combination strategies targeting both the primary tumor and disseminating cell states with host tissue commonalities, centering on bypass pathways to deter drug resistance and metastasis.
Amyloid fibrils are involved in devastating conditions such as Alzheimer's disease, Parkinson's disease, Huntington's disease, and systemic amyloidosis. They exhibit polymorphism, meaning that a single protein sequence can adopt different amyloid folds that vary with time and self-assembly conditions. Polymorphism confounds structure-based drug design and raises fundamental questions regarding why particular fibril structures form and how they cause disease. Here, we highlight the latest advances in our understanding of amyloid polymorphism, including its structural basis, thermodynamic origins, kinetic influences, and significance for disease. The next frontier will be to predict fibril structures, disentangle the dynamic mechanisms that guide the progression of fibril polymorphs, and illuminate how cofactors and the physiological milieu select for particular polymorphs in disease.
High-speed atomic force microscopy (HS-AFM) enables direct nanometer-resolution visualization of single molecules and molecular assemblies in real-time and under physiological conditions, providing unique insights into how membranes and membrane proteins move and interact within native lipid environments. Recent methodological advances and integration with complementary techniques have extended HS-AFM to increasingly complex, physiologically relevant systems, bridging gaps between high-resolution static structural methods and low-resolution functional dynamics. Here, we highlight how HS-AFM has changed our understanding of membrane organization, protein conformational dynamics, and lipid-protein coupling. By capturing transient events inaccessible to ensemble approaches, HS-AFM is transforming our ability to connect structural snapshots with functional behavior, advancing dynamic structural biology.