Biology subjects

Kolossvary, I.

Publications and source records attributed to Kolossvary, I..

6 recordsLinked to original sources

bioRxiv2026

Efficient detection of cryptic pockets on challenging drug targets using MD-free co-vibrational modes

Allosteric and cryptic binding sites provide unique opportunities in drug discovery for achieving selectivity, reducing toxicity and unlocking challenging therapeutical targets. Yet, cryptic pockets are particularly elusive in that they are difficult to capture with experimental methodologies, and therefore often remain undetected. Computational approaches for flexible cryptic site detection traditionally involve extensive unbiased molecular dynamics simulations, or more efficient but computationally still highly demanding biased MD alternatives that make use of enhanced sampling along low-frequency motions of the protein structure. Here we utilize a new variant of the recently introduced and orders of magnitude faster, MD-free COVIB algorithm to identify cryptic pockets efficiently in two challenging targets, the KRAS oncoprotein and the N-terminal domain of the STAT5B transcription factor. Using experimental and AI-modelled structures, respectively, we identify numerous cryptic sites, some of which have not yet been reported. The sites are prioritized and validated with our pharmacophore optimized photoaffinity-tagged fragment library via MS-based peptide mapping and NMR spectroscopy. Besides identifying new, targetable regions on the surface of these important oncotargets, we have established a protocol combining COVIB-based cryptic pocket detection and photoaffinity-based experimental peptide mapping as a robust approach for exploring unique and tractable cryptic pockets.

biophysics↗

bioRxiv2025

Interactions with lipid membrane modulate the conformational dynamics and energy landscape of Tumor Necrosis Factor

Tumor Necrosis Factor (TNF) is a trimeric cytokine that exists in soluble (sTNF) and membrane-bound (mTNF) forms, both of which regulate immune responses through interactions with their cognate receptors. sTNF has been the subject of extensive biophysical investigations, leading to a mechanistic model in which a symmetric arrangement of the trimer promotes receptor signaling, whereas asymmetric conformations inhibit this function. In contrast, the structural and energetic landscape of mTNF remains largely under-explored. Here, we combined multi-microsecond unbiased molecular dynamics and Metadynamics-Adaptive Biasing Force simulations to characterize mTNF embedded in a physiologically relevant lipid bilayer. We show that in the absence of membrane engagement, the extracellular domain (ECD) dynamically samples multiple asymmetric conformations similar to those observed in sTNF. The association of the ECD with the membrane, mediated primarily by the basic residues R78-R82, R107, R108, R120, and R207, restricts conformational heterogeneity and stabilizes the symmetric state. By quantifying the energetic effects of ECD-membrane association, we demonstrate that the symmetric state in the membrane-bound ECD is stabilized over asymmetric conformations to a greater extent than in sTNF. This energetic effect, together with the spatial confinement imposed by the lipid bilayer, may explain the previously reported reduction in affinity of certain biologics for mTNF. In conclusion, our work elucidates how the membrane influences the structure, dynamics, and energetics of TNF. These mechanistic insights could guide future efforts to design mTNF-selective inhibitors that account for both membrane constraints and the energetic modulation of the ECD conformational landscape.

biophysics↗

bioRxiv2025

Reinforced molecular dynamics: Physics-infused generative machine learning model explores CRBN activation process

We propose a simple and practical machine learning-based desktop solution for modeling biologically relevant protein motions. We termed our technology reinforced molecular dynamics (rMD) combining MD trajectory data and free-energy (FE) map data to train a dual-loss function autoencoder network that can explore conformational space more efficiently than the underlying MD simulation. The key insight of rMD is that it effectively replaces the latent space with an FE map, thus infusing the autoencoder network with a physical context. The FE map is computed from an MD simulation over a low-dimensional collective variable space that captures some biological function. One can directly use then the FE map for example, to generate more protein structures in poorly sampled regions, follow paths on the FE map to explore conformational transitions, etc. The rMD technology is entirely self-contained, does not rely on any pre-trained model, and can be run on a single GPU desktop computer. We present our rMD computations in a key area of molecular-glue degraders aimed at a deeper understanding of the structural transition from open to closed conformations of CRBN.

molecular biology↗

bioRxiv2022

Design of a systemic small molecule clinical STING agonist using physics-based simulations and artificial intelligence

The protein STING (stimulator of interferon genes) is a central regulator of the innate immune system and plays an important role in antitumor immunity by inducing the production of cytokines such as type I interferon (IFN). Activation of STING stems from the selective recognition of endogenous cyclic dinucleotides (CDNs) by the large, polar, and flexible binding site, thus posing challenges to the design of small molecule agonists with drug-like physicochemical properties. In this work we present the design of SNX281, a small molecule STING agonist that functions through a unique self-dimerizing mechanism in the STING binding site, where the ligand dimer approximates the size and shape of a cyclic dinucleotide while maintaining drug-like small molecule properties. SNX281 exhibits systemic exposure, STING-mediated cytokine release, strong induction of type I IFN, potent in vivo antitumor activity, durable immune memory, and single-dose tumor elimination in mouse models via a Cmax-driven pharmacologic response. Bespoke computational methods - a combination of quantum mechanics, molecular dynamics, binding free energy simulations, and artificial intelligence - were developed during the course of the project to design SNX281 by explicitly accounting for the unique self-dimerization mechanism and the large-scale conformational change of the STING protein upon activation. Over the course of the project, we explored millions of virtual molecules while synthesizing and testing only 208 molecules in the lab. This work highlights the value of a multifaceted computationally-driven approach anchored by methods tailored to address target-specific problems encountered along the project progression from initial hit to the clinic.

cancer biology↗

bioRxiv2021

Atomic-Resolution Prediction of Degrader-mediated Ternary Complex Structures by Combining Molecular Simulations with Hydrogen Deuterium Exchange

Targeted protein degradation (TPD) has emerged as a powerful approach in drug discovery for removing (rather than inhibiting) proteins implicated in diseases. A key step in this approach is the formation of an induced proximity complex, where a degrader molecule recruits an E3 ligase to the protein of interest (POI), facilitating the transfer of ubiquitin to the POI and initiating the proteasomal degradation process. Here, we address three critical aspects of the TPD process: 1) formation of the ternary complex induced by a degrader molecule, 2) conformational heterogeneity of the ternary complex, and 3) assessment of ubiquitination propensity via the full Cullin Ring Ligase (CRL) macromolecular assembly. The novel approach presented here combines experimental biophysical data--in this case hydrogen-deuterium exchange mass spectrometry (HDX-MS, which measures the solvent exposure of protein residues)--with all-atom explicit solvent molecular dynamics (MD) simulations aided by enhanced sampling techniques to predict structural ensembles of ternary complexes at atomic resolution. We present results demonstrating the efficiency, accuracy, and reliability of our approach to predict ternary structure ensembles using the bromodomain of SMARCA2 (SMARCA2BD) with the E3 ligase VHL as the system of interest. The simulations reproduce X-ray crystal structures - including prospective simulations validated on a new structure that we determined in this work (PDB ID: 7S4E) - with root mean square deviations (RMSD) of 1.1 to 1.6 [A]. The simulations also reveal a structural ensemble of low-energy conformations of the ternary complex within a broad energy basin. To further characterize the structural ensemble, we used snapshots from the aforementioned simulations as seeds for Hamiltonian replica exchange molecular dynamics (HREMD) simulations, and then perform 7.1 milliseconds of aggregate simulation time using Folding@home. The resulting free energy surface identifies the crystal structure conformation within a broad low-energy basin and the dynamic ensemble is consistent with solution-phase biophysical experimental data (HDX-MS and small-angle x-ray scattering, SAXS). Finally, we graft structures from the ternary complexes onto the full CRL and perform enhanced sampling simulations, where we find that differences in degradation efficiency can be explained by the proximity distribution of lysine residues on the POI relative to the E2-loaded ubiquitin. Several of the top predicted ubiquitinated lysine residues are validated prospectively through a ubiquitin mapping proteomics experiment.

biophysics↗

bioRxiv2020

The N34S mutation of SPINK1 may impact the kinetics of trypsinogen activation to cause early trypsin release in the pancreas

The N34S variant of the trypsin inhibitor SPINK1 is the clinically most significant risk factor for chronic pancreatitis, but the underlying molecular mechanism could not be identified. Molecular dynamics simulations and docking of the generated conformational ensemble of SPINK1 to trypsin show that the mutation reduces the fraction of conformations that can directly participate in productive association, thereby reducing the association rate. The small change is difficult to detect by measuring the kinetics of SPINK1 binding to trypsin. However, kinetic modeling reveals that even a small change in the inhibition rate affects the trypsinogen to trypsin conversion rate at the early stage of the reaction when the trypsin concentration is very low, and the impact is substantially amplified by the autocatalytic mechanism of the conversion. Thus, the slightly reduced inhibition rate shortens the delay in the activation of trypsin release, which is therefore occurs within the pancreas.

biophysics↗