Zhongjie Liang

Missense mutations and post-translational modifications (PTMs) are major molecular perturbations that reshape protein function but are traditionally studied independently. Current computational approaches largely rely on sequence conservation or static structural features, limiting our understanding of how perturbations alter intrinsic protein dynamics. We present DynGeo-Pheno, a unified geometric deep learning framework that integrates protein language model representations with anisotropic network model-derived dynamics to jointly capture evolutionary, structural, and biophysical information. DynGeo-Pheno predicts disease-associated phosphosites and pathogenic missense mutations with high accuracy on independent test datasets. Ablation analyses indicate that protein dynamics provide complementary information beyond sequence evolution and structural topology for pathogenicity prediction. Beyond predictive performance, DynGeo-Pheno reveals that disease-associated perturbations preferentially localize to functional structural regions, including ligand-binding pockets and PPI interfaces. Mechanistically, phosphosites and missense mutations appear to exhibit distinct yet convergent dynamic signatures. Phosphosites preferentially occur in flexible regulatory regions, whereas pathogenic mutations are enriched in ordered structural elements. Despite these differences, both perturbation types display enhanced long-range coupling, increased perturbation responsiveness, and elevated mechanical stability, indicating that pathogenic residues preferentially occupy mechanically constrained and allosteric regulatory sites. This study provides compelling evidence that intrinsic protein dynamics is an important complementary determinant of pathogenicity and establishes a unified framework for interpretable AI predictions and mechanistic understanding of how genetic and regulatory perturbations may shape protein function.

Post-translational modifications (PTMs) and somatic mutations are pervasive in cancer, yet their three-dimensional organization and coordinated regulatory mechanisms within protein–protein interactions (PPIs) remain poorly understood. Here we develop ClusTar, a structural framework that systematically identifies spatial clusters of PTMs and mutations within PPIs and links them to protein dynamics and functional regulation. Integrating TCGA genomics, CPTAC proteomics, and structural PPI data across ten cancer types, we identify 3,966 high-confidence PTM–mutation clusters that are enriched in functional residues and frequently overlap with ligand-binding pockets. Dynamics analyses reveal that many clusters are mechanically coupled to PPI interfaces and mediate long-range allosteric communications. In RhoA complexes, glutamine substitutions at acetylated lysine sites (K7Q, K18Q, K104Q and K118Q/K162Q) have been observed to redistribute key network hubs and rewire allosteric communication pathways among clustered residues. Experimental validation demonstrates that these mutations strengthen protein interactions and promote breast cancer cell migration, highlighting their potential as allosteric hotspots. Together, our results establish a pan-cancer structural atlas of PTM–mutation clusters and highlight allosteric hotspots as potential targets for modulating PPIs.

Pathogenic mutations in “undruggable” Ras superfamily proteins challenge drug development by inducing subtle, dynamic conformational changes. Here, we integrated X-ray crystallography with native mass spectrometry and ultraviolet photodissociation (nMS-UVPD) to reveal a cryptic conformation in the oncogenic Y42C mutant of RhoA. While crystallography alone resolved two ambiguous structures, nMS-UVPD determined the dominant conformation by directly mapping the mutant’s conformational dynamics, identifying an enhanced Mg²⁺-locked conformation. We explored the mechanism of mutation impairing GTP hydrolysis. This state unmasks a previously hidden, druggable pocket adjacent to Cys42, guiding our identification of a covalent inhibitor. Our integrated approach establishes a roadmap for targeting pathogenic protein mutants previously considered “undruggable” due to their highly dynamic nature.

Phosphorylation dynamically orchestrates the protein-protein interaction (PPI) network that governs cellular signaling, and its dysregulation frequently drives malignant transformation and neurodegeneration. We present PhosPPI-SEQ, an interpretable deep learning framework to identify phosphosites that functionally regulate PPIs. The model integrates a pretrained Transformer architecture with a cross-attention mechanism for fine-tuning, achieving both high predictive accuracy and biologically meaningful attention patterns. Combining PhosPPI-SEQ with a regression model based on pan-cancer multiomics data, we map context-specific phosphosites involved in PPI regulation across six cancer types, which are validated experimentally in the context of epigenetic regulation. Notably, hyperphosphorylation of HDAC1 at S421 or S423 strengthens its binding to KDM1A and correlates with poor clinical outcomes in small-cell lung cancer (SCLC) patients. A dual inhibitor targeting class I HDAC and KDM1A exhibits superior antiproliferative effects on SCLC cells expressing HDAC1 S421 or S423 phosphomimetic mutations. Our study establishes a computational-experimental paradigm for decoding phosphosites that regulate PPIs, significantly expanding the targetable PPI landscape in a phosphosite biomarker-oriented manner.

The bromodomain (BRD) represents a highly conserved structural module that provides BRD proteins with fundamental functionality in modulating protein-protein interactions involved in diverse biological processes such as chromatin-mediated gene transcription, DNA recombination, replication and repair. Consequently, dysregulation of BRD proteins has been implicated in the pathogenesis of numerous human diseases. In recent years, considerable scientific endeavors have focused on unraveling the molecular mechanisms underlying BRDs and developing inhibitors that target these domains. While these inhibitors compete for binding with the acetylated lysine binding site of BRDs, achieving inhibition of BRD proteins via competitive pocket binding has proven challenging due to the conserved nature of these pockets. To address this limitation, the present study employed dynamic simulations for a comprehensive analysis, leading to the identification of a non-conserved pocket in CECR2 for achieving BRD family inhibition through allosteric modulation. Subsequently, the compound BAY 11-7085 was proven capable of covalently binding to C494 of this pocket after covalent docking and biological verification in vitro. The allosteric inhibition strategy of CECR2 was further verified by the structurally optimized compound LC-CE-7, which is an allosteric covalent CECR2 inhibitor with anti-cancer effects in MDA-MB-231 cells.

Post-translational modifications (PTMs) fine-tune various signaling pathways not only by the modification of a single residue, but also by the interplay of different modifications on residue pairs within or between proteins, defined as PTM cross-talk. As a challenging question, less attention has been given to PTM dynamics underlying cross-talk residue pairs and structural information underlying protein–protein interaction (PPI) graph, limiting the progress in this PTM functional research. Here we propose a novel integrated deep neural network PPICT (Predictor for PTM Inter-protein Cross-Talk), which predicts PTM cross-talk by combining protein sequence-structure-dynamics information and structural information for PPI graph. We find that cross-talk events preferentially occur among residues with high co-evolution and high potential in allosteric regulation. To make full use of the complex associations between protein evolutionary and biophysical features, and protein pair features, a heterogeneous feature combination net is introduced in the final prediction of PPICT. The comprehensive test results show that the proposed PPICT method significantly improves the prediction performance with an AUC value of 0.869, outperforming the existing state-of-the-art methods. Additionally, the PPICT method can capture the potential PTM cross-talks involved in the functional regulatory PTMs on modifying enzymes and their catalyzed PTM substrates. Therefore, PPICT represents an effective tool for identifying PTM cross-talk between proteins at the proteome level and highlights the hints for cross-talk between different signal pathways introduced by PTMs.