ABSTRACT Accurate prediction of nucleic acid‐binding sites is essential for understanding protein function and facilitating drug discovery. However, current computational methods are limited by single‐layer language model representations, Cα point clouds that discard local‐frame geometry, and feature fusion in which high‐dimensional sequence embeddings overshadow weaker evolutionary and structural signals. Here, we present PNBind, a multimodal deep learning framework that leverages protein language models, evolutionary information, and three‐dimensional structural features via a geometric vector perceptron (GVP) backbone with an invariant point attention (IPA) layer for nucleic acid‐binding site prediction. By integrating sequence representations from the ESM‐series language models with alignment‐derived evolutionary features, secondary‐structure data, and geometric features extracted from five‐atom local frames, PNBind enables effective late‐stage multimodal signal fusion. Comprehensive benchmarking on the DNA‐129, DNA‐181, RNA‐117, and RNA‐285 datasets reveals that PNBind achieves strong performance, with F 1 scores of 0.611, 0.443, 0.370, and 0.472 and MCC values of 0.586, 0.425, 0.351, and 0.416, respectively. Ablation studies confirm the dominant contribution of the protein language model backbone, while systematic evaluation of ESM‐family protein language models demonstrates that proper PLM selection significantly boosts prediction accuracy. This study demonstrates that the effective integration of protein language models and geometric deep learning offers a practical approach for accurate nucleic acid‐binding site prediction.
Recent advances in protein structure prediction have considerably increased the number of available structures, underscoring the need for scalable and accurate methods to compare and classify this huge amount of data. Current approaches based on sequence alignment or structural superposition often lack sensitivity and/or are computationally expensive. Here, we introduce the DeltaFold Classifier (DFC), a fast, alignment-free, protein structure classification pipeline based on topological data analysis. Protein three-dimensional structures are encoded using fixed-length vectors derived from persistent homology applied to point clouds of their spatial representation. These vectors, known as Biotopological Markers (BTMs), capture the intrinsic topological features of protein structure geometry-such as cycles and cavities-across multiple spatial scales. BTMs are invariant under translations and rotations, thereby avoiding the need for structural superposition. The DFC pipeline uses the features of these BTMs to train machine learning models for protein structure classification at various hierarchical levels defined in the SCOP and CATH classifications. It achieves performance comparable to that of structure-based comparison methods while substantially improving computational efficiency. It also outperforms sequence-based methods in tasks involving distant homology detection. In-depth analyses confirm that most BTM features contribute significantly to classification performance. These findings highlight the potential of topological vectorisations in structural bioinformatics and support the DFC pipeline as an effective and scalable tool for automated protein structure classification and annotation.
Protein Tyrosine Phosphatases (PTPs) regulate cellular signaling by balancing phosphotyrosine levels through a conserved WPD-loop that switches between open (inactive) and closed (active) conformations. While this conformational change is crucial for catalysis, the structural factors governing this motion remain poorly understood. We analyzed 551 PTP crystal structures using a modified Local Spatial Patterns (LSP) algorithm and XGBoost machine learning to identify core residues whose sidechain conformations influence WPD-loop dynamics. Our analysis revealed two distinct residue clusters: one centered on F225 (PTP1B numbering) in the protein core and another on the α2 helix approximately 20 Å from the WPD-loop. Remarkably, we discovered subfamily-specific regulatory mechanisms involving the conserved phenylalanine at position 225. Classical PTPs like PTP1B maintain this residue predominantly in a "down" conformation regardless of WPD-loop state, while SHP2 exhibits unique conformational flexibility with its equivalent residue F469 displaying dual occupancy between "up" and "down" conformations. This reveals that different PTP subfamilies evolved distinct allosteric control mechanisms while maintaining the overall catalytic framework. These subfamily-specific conformational switches have significant therapeutic implications. PTP1B's role as a negative insulin signaling regulator makes it an attractive diabetes target, while SHP2's function in growth factor signaling and oncogenesis suggests cancer therapeutic opportunities. Our results provide a structural framework for understanding subfamily-specific WPD-loop regulation and identify sites for selective therapeutic intervention, offering strategies for developing highly selective inhibitors that circumvent traditional active site-directed approach limitations.
The E2 domain of amyloid precursor protein (APP_E2) plays a central role in the protein's function. We reinvestigate how the missense mutation, ΔH382L, affects the domain's thermal stability and ligand binding propensity through differential scanning fluorometry. We report that the ΔH382L significantly impacts the thermal stability of APP_E2 with a reduction in Tm of 10.1°C ± 0.8°C when compared to wild-type APP_E2. In the presence of Cu2+ ions, WT APP_E2 showed a significant increase in Tm of 2.3°C ± 0.4°C, indicative of binding, whereas the mutant exhibited a significant decrease in Tm of 9.5°C ± 0.4°C, suggesting that Cu2+-mediated stabilization of APP_E2 is highly dependent on specific interactions. We further observed that the mutation affected heparin binding, with the mutant protein showing a 9.9°C ± 0.7°C decrease in Tm and requiring lower NaCl concentrations for elution from heparin Sepharose compared with the wild-type protein. Lastly, through molecular dynamics simulations, native contacts, dynamic cross-correlation and network analyses, we demonstrated that this point mutation may induce rigidity to one of the helical subdomains of APP_E2, change native contacts and correlated motion across the domain and reroute possible established communication pathways from the mutation site to distal residues. The global perturbations predicted in the presence of mutation may provide possible mechanistic explanation of the observed reduced thermal tolerance. Overall, our data reported that a missense mutation to an uncharged amino acid at position 382 of APP_E2 significantly decreases the thermal stability and binding propensity of the domain.
An advanced theory based on statistical physics was applied to microscopically investigate androstenone adsorption and its relationship with olfactory perception in chimpanzees and gorillas. For this purpose, dose-response curves of androstenone on olfactory receptors of chimpanzee OR7D4 and gorilla OR7D4 were analyzed by statistical physics. The two-energy adsorption model fits the chimpanzee data well, and the gorilla data are best fit by the one-energy adsorption model. The stereographic parameter analysis indicated that the studied pheromones were docked in a nonparallel manner in both species via a multimolecular mechanism. The analysis showed that the molecules interacted with different amino acid residues, resulting in distinct olfactory responses. The positive values of the molar adsorption energies indicate that an exothermic and physisorption process occurred in both olfactory systems. The docking analysis revealed that androstenone binds to the olfactory receptors of chimpanzees and gorillas through specific interactions with amino acid residues in the binding pockets. The docking results indicated distinct binding modes in each species, involving van der Waals, alkyl, and C-H interactions, which corroborated the energy trends predicted by the statistical physics model. These findings provided insights into the molecular mechanisms underlying the differential olfactory responses in chimpanzees and gorillas. Statistical physics was an excellent tool for evaluating and interpreting the interactions between androstenone and olfactory receptors.
