GATSol, an enhanced predictor of protein solubility through the synergy of 3D structure graph and large language modeling.

BACKGROUND: Protein solubility is a critically important physicochemical property closely related to protein expression. For example, it is one of the main factors to be considered in the design and production of antibody drugs and a prerequisite for realizing various protein functions. Although several solubility prediction models have emerged in recent years, many of these models are limited to capturing information embedded in one-dimensional amino acid sequences, resulting in unsatisfactory predictive performance. RESULTS: In this study, we introduce a novel Graph Attention network-based protein Solubility model, GATSol, which represents the 3D structure of proteins as a protein graph. In addition to the node features of amino acids extracted by the state-of-the-art protein large language model, GATSol utilizes amino acid distance maps generated using the latest AlphaFold technology. Rigorous testing on independent eSOL and the Saccharomyces cerevisiae test datasets has shown that GATSol outperforms most recently introduced models, especially with respect to the coefficient of determination R2, which reaches 0.517 and 0.424, respectively. It outperforms the current state-of-the-art GraphSol by 18.4% on the S. cerevisiae_test set. CONCLUSIONS: GATSol captures 3D dimensional features of proteins by building protein graphs, which significantly improves the accuracy of protein solubility prediction. Recent advances in protein structure modeling allow our method to incorporate spatial structure features extracted from predicted structures into the model by relying only on the input of protein sequences, which simplifies the entire graph neural network prediction process, making it more user-friendly and efficient. As a result, GATSol may help prioritize highly soluble proteins, ultimately reducing the cost and effort of experimental work. The source code and data of the GATSol model are freely available at https://github.com/binbinbinv/GATSol .

How alginate lyase produces quasi-monodisperse oligosaccharides: A normal-mode-based docking and molecular dynamics simulation study.

Oligosaccharide degradation products of alginate (AOS) hold significant potential in diverse fields, including pharmaceuticals, health foods, textiles, and agricultural production. Enzymatic alginate degradation is appealing due to its mild conditions, predictable activity, high yields, and controllability. However, the alginate degradation often results in a complex mixture of oligosaccharides, necessitating costly purification to isolate highly active oligosaccharides with a specific degree of polymerization (DP). Addressing this, our study centers on the alginate lyase AlyB from Vibrio Splendidus OU02, which uniquely breaks down alginate into mono-distributed trisaccharides. This enzyme features a polysaccharide lyase family 7 domain (PL-7) and a CBM32 carbohydrate-binding module connected by a helical structure. Through normal-mode-based docking and all-atom molecular simulations, we demonstrate that AlyB's substrate and product specificities are influenced by the spatial conformation of the catalytic pocket and the flexibility of its structure. The helically attached CBM is pivotal in releasing trisaccharides, which is crucial for avoiding further degradation. This study sheds light on AlyB's specificity and efficiency and contributes to the evolving field of enzyme design for producing targeted oligosaccharides, with significant implications for various bioindustries.

A Comprehensive Self-Resistance Gene Database for Natural-Product Discovery with an Application to Marine Bacterial Genome Mining.

In the world of microorganisms, the biosynthesis of natural products in secondary metabolism and the self-resistance of the host always occur together and complement each other. Identifying resistance genes from biosynthetic gene clusters (BGCs) helps us understand the self-defense mechanism and predict the biological activity of natural products synthesized by microorganisms. However, a comprehensive database of resistance genes is still lacking, which hinders natural product annotation studies in large-scale genome mining. In this study, we compiled a resistance gene database (RGDB) by scanning the four available databases: CARD, MIBiG, NCBIAMR, and UniProt. Every resistance gene in the database was annotated with resistance mechanisms and possibly involved chemical compounds, using manual annotation and transformation from the resource databases. The RGDB was applied to analyze resistance genes in 7432 BGCs in 1390 genomes from a marine microbiome project. Our calculation showed that the RGDB successfully identified resistance genes for more than half of the BGCs, suggesting that the database helps prioritize BGCs that produce biologically active natural products.

Detection and treatment of biofilm-induced periodontitis by histidine-doped FeSN nanozyme with ultra-high peroxidase-like activity.

Pathogenic biofilm induced oral diseases have posed a significant treat to human health, such as periodontitis resulting from the formation of bacterial biofilm on teeth and gums. The traditional treatment methods such as mechanical debridement and antibiotic therapy encounter the poor therapeutic effect. Recently, numerous nanozymes with excellent antibacterial effect have been widely used in the treatment of oral diseases. In this study, a novel iron-based nanozyme (FeSN) generated by histidine-doped FeS2 with high peroxidase-like (POD-like) activity was designed for the oral biofilm removal and treatment of periodontitis. FeSN exhibited an extremely high POD-like activity, and enzymatic reaction kinetics and theoretical calculations had demonstrated its catalytic efficiency to be approximately 30 times than that of FeS2. The antibacterial experiments showed that FeSN had robust antibacterial activity against Fusobacterium nucleatum in the presence of H2O2, causing a reduction in the levels of glutathione reductase and ATP in bacterial cells, while increasing the level of oxidase coenzyme. The ultrahigh POD-like activity of FeSN allowed for easy detection of pathogenic biofilms and promoted the breakdown of biofilm structure. Furthermore, FeSN demonstrated excellent biocompatibility and low cytotoxicity to human fibroblast cells. In a rat model of periodontitis, FeSN exhibited significant therapeutic effects by reducing the extent of biofilm formation, inflammation, and alveolar bone loss. Taken together, our results suggested that FeSN, generated by self-assembly of two amino acids, represented a promising approach for biofilm removal and periodontitis treatment. This method has the potential to overcome the limitations of current treatments and provide an effective alternative for periodontitis treatment.

Highly efficiency production of D-allulose from inulin using curli fiber multi-enzyme cascade catalysis.

D-Allulose is a rare sugar with numerous physiological benefits and low caloric content, which can be obtained from inulin through enzymatic catalysis. In this study, we combined D-allulose 3-epimerase, exo- and endo-inulinases (EXINU and ENINU) with the NGTag/NGCatcher/CsgA system to accelerate D-allulose accumulation from inulin. Molecular dynamics simulations were used to screen linkers of appropriate length for ENINU. In vitro, we successfully observed the assembled NGCatcher_ENINU_CsgA, NGTag_EXINU, and DAERK fibers using fluorescent labelling with GFP, YFP, and mCherry. The optimal pH and temperature of the tagged variants were comparable to those of the wild-type, and the MD simulations showed that NGCatcher_ENINU_CsgA had improved stability in the working environment of EXINU. D-Allulose accumulation rate of the assembled enzymes cascade (NGCatcher_ENINU_CsgA/NGTag_EXINU_CsgA/DAERK) reached 0.25 g/L min-1 (1.25 mgD-allulose mgDAERK-1 min-1) at an inulin concentration of 100 g/L. The assembled system greatly improves the high-valued productions of rare sugars from cheap biomass.

Ratiometric fluorescence probe integrated with smartphone for visually detecting lipopolysaccharide.

A portable instrument-free detection method for lipopolysaccharide (LPS) analysis was developed based on dual-emission ratiometric fluorescence sensing system. Herein, red-emitting Au nanoclusters (Au NCs) were as reference probe, while blue-emitting fluorescent silica quantum dots (Si QDs) were as response probe. Additionally, the aptamer of LPS was covalently grafted to the surface of Si QDs in order to specific recognize the LPS. According to the changes of fluorescence intensityratio (FL ratio, I461 nm/I643 nm) with the concentrations of LPS, the linear equation was fitted with the range of 50-3000 ng/mL, and the limit of detection (LOD) was 29.3 ng/mL. As a practical application, this method was employed to analyze LPS in normal saline with the recovery rate of 97.7-103.8 %. The color picker platform in the smartphone was used to transform the detection picture to the process of Red, Green and Blue (RGB) for visual detection of LPS. The low-cost and easy-carry method reported here presents broad merits for the visually quantitative detection of LPS.

A multilayer dynamic perturbation analysis method for predicting ligand-protein interactions.

BACKGROUND: Ligand-protein interactions play a key role in defining protein function, and detecting natural ligands for a given protein is thus a very important bioengineering task. In particular, with the rapid development of AI-based structure prediction algorithms, batch structural models with high reliability and accuracy can be obtained at low cost, giving rise to the urgent requirement for the prediction of natural ligands based on protein structures. In recent years, although several structure-based methods have been developed to predict ligand-binding pockets and ligand-binding sites, accurate and rapid methods are still lacking, especially for the prediction of ligand-binding regions and the spatial extension of ligands in the pockets. RESULTS: In this paper, we proposed a multilayer dynamics perturbation analysis (MDPA) method for predicting ligand-binding regions based solely on protein structure, which is an extended version of our previously developed fast dynamic perturbation analysis (FDPA) method. In MDPA/FDPA, ligand binding tends to occur in regions that cause large changes in protein conformational dynamics. MDPA, examined using a standard validation dataset of ligand-protein complexes, yielded an averaged ligand-binding site prediction Matthews coefficient of 0.40, with a prediction precision of at least 50% for 71% of the cases. In particular, for 80% of the cases, the predicted ligand-binding region overlaps the natural ligand by at least 50%. The method was also compared with other state-of-the-art structure-based methods. CONCLUSIONS: MDPA is a structure-based method to detect ligand-binding regions on protein surface. Our calculations suggested that a range of spaces inside the protein pockets has subtle interactions with the protein, which can significantly impact on the overall dynamics of the protein. This work provides a valuable tool as a starting point upon which further docking and analysis methods can be used for natural ligand detection in protein functional annotation. The source code of MDPA method is freely available at: https://github.com/mingdengming/mdpa .

Tailoring the Tag/Catcher System by Integrating Covalent Bonds and Noncovalent Interactions for Highly Efficient Protein Self-Assembly.

Covalent bonds and noncovalent interactions play crucial roles in enzyme self-assembly. Here, we designed a Tag/Catcher system named NGTag/NGCatcher in which the Catcher is a highly charged protein that can bind proteins with positively charged tails and rapidly form a stable isopeptide bond with NGTag. In this study, we present a multienzyme strategy based on covalent bonds and noncovalent interactions. In vitro, mCherry, YFP, and GFP can form protein-rich three-dimensional networks based on NGCatcher, NGTag, and RK (Arginine/Lysine) tails, respectively. Furthermore, this technology was applied to improve lycopene production in Escherichia coli. Three key enzymes were involved in lycopene production variants from Deinococcus wulumuqiensis R12 of NGCatcher_CrtE, NGTag_Idi, and RKIspARK, where the multienzyme complexes were clearly observed in vivo and in vitro, and the lycopene production in vivo was 17.8-fold higher than that in the control group. The NGTag/NGCatcher system will provide new opportunities for in vivo and in vitro multienzyme catalysis.

Sequence, structure, and function of the Dps DNA-binding protein from Deinococcus wulumuqiensis R12.

Deinococcus wulumuqiensis R12, which was isolated from arid irradiated soil in Xinjiang province of China, belongs to a genus that is well-known for its extreme resistance to ionizing radiation and oxidative stress. The DNA-binding protein Dps has been studied for its great contribution to oxidative resistance. To explore the role of Dps in D. wulumuqiensis R12, the Dps sequence and homology-modeled structure were analyzed. In addition, the dps gene was knocked out and proteomics was used to verify the functions of Dps in D. wulumuqiensis R12. Docking data and DNA binding experiments in vitro showed that the R12 Dps protein has a better DNA binding ability than the Dps1 protein from D. radiodurans R1. When the dps gene was deleted in D. wulumuqiensis R12, its resistance to H2O2 and UV rays was greatly reduced, and the cell envelope was destroyed by H2O2 treatment. Additionally, the qRT-PCR and proteomics data suggested that when the dps gene was deleted, the catalase gene was significantly down-regulated. The proteomics data indicated that the metabolism, transport and oxidation-reduction processes of D. wulumuqiensis R12 were down-regulated after the deletion of the dps gene. Overall, the data conformed that Dps protein plays an important role in D. wulumuqiensis R12.

Near-Infrared-Fluorescent Probe for Turn-On Lipopolysaccharide Analysis Based on PEG-Modified Gold Nanorods with Plasmon-Enhanced Fluorescence.

Lipopolysaccharide (LPS), as the major component of the outer membrane of Gram-negative bacteria, can trigger a variety of biological effects such as sepsis, septic shock, and even multiorgan failure. Herein, we developed a near-infrared-fluorescent probe for fluorescent turn-on analysis of LPS based on plasmon-enhanced fluorescence (PEF). Gold nanorods (Au NRs) modified polyethylene glycol (PEG) was used as PEF materials. Au NRs were prepared with different longitudinal surface plasmon resonance (LSPR), and their fluorescence enhancement was investigated. Three kinds of molecular weights (1000, 5000, and 10000) of polyethylene glycol (PEG) were employed to control the distance between the Au NRs and the fluorescence substances of cyanine 7 (Cy7). Experimental analysis showed that the enhancement was related to the spectral overlap between the plasmon resonance of Au NRs and the extinction/emission of fluorophore. The three-dimensional finite-difference time-domain (3D-FDTD) simulation further revealed that the enhancement was caused by local electric field enhancement. Furthermore, the probe was used for the ultrasensitive analysis of LPS with a detection limit of 3.85 ng/mL and could quickly distinguish the Gram-negative bacterium-Escherichia coli (E. coli) (with LPS in the membrane) from Gram-positive bacterium-Staphylococcus aureus (S. aureus) (without LPS), as well as quantitative determination of E. coli with a detection limit of 1.0 × 106 cfu/mL. These results suggested that the prepared probe has great potential for biomedical diagnosis and selective detection of LPS from different bacterial strains.

A switchable and signal-amplified aptasensor based on metal organic frameworks as the quencher for turn-on detection of T-2 mycotoxin.

A simple and low-cost fluorescence aptasensor was developed for rapid and sensitive signal amplification detection of T-2 mycotoxin (T-2). Dual-terminal-fluorescein amidite (FAM)-labeled aptamer (D-aptamer) acted as a recognition element and signal indicator. The metal organic frameworks (MOFs) of N, N'-bis(2-hydroxyethyl)dithiooxamidato copper (II) (H2dtoaCu) were as the quencher. The D-aptamer was initially adsorbed to the surface of H2dtoaCu, leading to efficient quenching of the aptasensor. Upon addition of T-2, the D-aptamer underwent a conformation change to form the T-2/T-2 aptamer complex, which induced the signaling probe to be released from the H2dtoaCu surface. Thus, the fluorescence intensity (FL) of the D-aptamer was recovered. Versus the single-terminal-FAM-labeled aptamer (S-aptamer), the D-aptamer showed a lower detection limit of 0.39 ng/mL. The aptasensor was also successfully applied to detect T-2 in corn and wheat samples with good recoveries.

Improving the thermostability of trehalose synthase from Thermomonospora curvata by covalent cyclization using peptide tags and investigation of the underlying molecular mechanism.

One of the most desirable properties for industrial enzymes is high thermotolerance, which can reduce the amount of biocatalyst used and lower the production cost. Aiming to improve the thermotolerance of trehalose synthase (TreS, EC 5.4.99.16) from Thermomonospora curvata, four mutants (G78D, V289L, G322A, I323L) and four cyclized TreS variants fused using different Tag/Catcher pairs (SpyTag-TreS-SpyCatcher, SpyTag-TreS-KTag, SnoopTag-TreS-SnoopCatcher, SnoopTagJR-TreS-DogTag) were constructed. The results showed that cyclization led to a much larger increase of thermostability than that achieved via site-directed mutagenesis. The t1/2 of all four cyclized TreS variants at 55 °C increased 2- to 3- fold, while the analysis of kinetic and thermodynamic stability indicated that the T50 of the different cyclized TreS variants increased by between 7.5 °C and 15.5 °C. Molecular dynamics simulations showed that the Rg values of cyclized TreS decreased significantly, indicating that the protein maintained a tight tertiary structure at high temperatures, avoiding exposure of the hydrophobic core to the solvent. Cyclization using a Tag/Catcher pair is a simple and effective method for improving the thermotolerance of enzymes.

Poly(N-Isopropylacrylamide)-Functional Silicon Nanocrystals for Thermosensitive Fluorescence Cellar Imaging.

Silicon nanocrystals (Si NCs) have received surging interest as a type of quantum dot (QD) due to the availability of silicon in nature, tunable fluorescence emission properties and excellent biocompatibility. More importantly, compared with many group II-VI and III-V based QDs, they have low toxicity. Here, thermoresponsive poly(N-isopropylacrylamide) (PNIPAAm)-functional Si NCs were firstly prepared for thermoresponsive detection of cancer cells. Si NCs were prepared under normal pressure with excellent water solubility. Then folic acid was bonded to the silicon nanocrystals through the reaction of amino and carboxyl groups for specific recognition of cancer cells. The folic-acid-modified silicon crystals (Si NCs-FA) could be modified by a one-pot copolymerization process into PNIPAAm nanospheres during the monomer polymerization process (i.e., Si NCs-FA-PNIPAAm) just by controlling the temperature below the lower critical solution temperature (LCST) and above the LCST. The results showed that the Si-FA-PNIAAm nanospheres exhibited not only reversible temperature-responsive on-off fluorescence properties, but also can be used as temperature indicators in cancer cells.

Whole-Genome Sequencing and Bioinformatics Analysis of Apiotrichum mycotoxinivorans: Predicting Putative Zearalenone-Degradation Enzymes.

Biological detoxification techniques have been developed by using microorganisms such as bacteria, yeast, and fungi to eliminate mycotoxin contamination. However, due to the lack of molecular details of related enzymes, the underlying mechanism of detoxification of many mycotoxins remain unclear. On the other hand, the next generation sequencing technology provides a large number of genomic data of microorganisms that can degrade mycotoxins, which makes it possible to use bioinformatics technology to study the molecular details of relevant enzymes. In this paper, we report the whole-genome sequencing of Apiotrichum mycotoxinivorans (Trichosporon mycotoxinivorans in old taxonomy) and the putative Baeyer-Villiger monooxygenases (BVMOs) and carboxylester hydrolases for zearalenone (ZEA) degradation through bioinformatic analysis. In particular, we developed a working pipeline for genome-scaled prediction of substrate-specific enzyme (GPSE, available at https://github.com/JinyuanSun/GPSE), which ultimately builds homologous structural and molecular docking models to demonstrate how the relevant degrading enzymes work. We expect that the enzyme-prediction woroflow process GPSE developed in this study might help accelerate the discovery of new detoxification enzymes.

The adaptability of H9N2 avian influenza A virus to humans: A comparative docking simulation study.

Influenza A virus, the H9N2 subtype, is an avian influenza virus that has long been circulating in the worldwide poultry industry and is occasionally found to be transmissible to humans. Evidence from genomic analysis suggests that H9N2 provides the genes for the H5N1 and H7N9 subtypes, which have been found to infect mammals and pose a threat to human health. However, due to the lack of a structural model of the interaction between H9N2 and host cells, the mechanism of the extensive adaptability and strong transformation capacity of H9N2 is not fully understood. In this paper, we collected 40 representative H9N2 virus samples reported recently, mainly in China and neighboring countries, and investigated the interactions between H9N2 hemagglutinin and the mammalian receptor, the polysaccharide α-2,6-linked lactoseries tetrasaccharide c, at the atomic level using docking simulation tools. We categorized the mutations of studied H9N2 hemagglutinin according to their effects on ligand-binding interactions and the phylogenetic analysis. The calculations indicated that all the studied H9N2 viruses can establish a tight binding with LSTc although the mutations caused a variety of perturbations to the local conformation of the binding pocket. Our calculations suggested that a marginal equilibrium is established between the conservative ligand-receptor interaction and the conformational dynamics of the binding pocket, and it might be this equilibrium that allows the virus to accommodate mutations to adapt to a variety of environments. Our results provided a way to understand the adaptive mechanisms of H9N2 viruses, which may help predict its propensity to spread in mammals.

Enhanced imaging of glycan expressing cancer cells using poly(glycidyl methacrylate)-grafted silica nanospheres labeled with quantum dots.

Glycosylation on the cell surface contains abundant biological information, and detecting the glycan on cell surfaces can offer critical insight into biology and diseases. Here, a signal amplification strategy for the sensitive detection of glycan expression on the cell surface was proposed. In this approach, glycans on the cell surface were detected with poly(glycidyl methacrylate)-grafted silica nanosphere labeled with quantum dots (QDs) and biotin through the specific affinity reaction of avidin-biotin on the cancer cells. Glycans on the cell surface were first labeled via selective oxidization of sialyl groups into aldehydes by periodate. Aniline-catalyzed hydrazone ligation with biotin hydrazide was then used for the specific recognition to avidin. The nanoprobe was fabricated with "living" SiO2 nanoparticles with alkyl bromide groups on their surfaces. They were then subsequently grafted with poly(glycidyl methacrylate) (PGMA) brushes via the successive surface-initiated atom transfer radical polymerization. The CdTe QDs and biotin were immobilized through a ring-open reaction with epoxy groups in the PGMA brushes to obtain QDs/biotin-polymer brush-functionalized silica nanosphere (SiO2-PGMA-QDs/biotin). Enhanced sensitivity could be achieved by an increase in CdTe QDs loading per assay event, because of the large number of surface functional epoxy groups offered by the PGMA. As a result, fluorescence signal increased versus the unamplified method. This method successfully demonstrates a simple, specific, and potent method to detect glycans on the cell surface.

Elucidation of a Unique Pattern and the Role of Carbohydrate Binding Module of an Alginate Lyase.

Alginate oligosaccharides with different degrees of polymerization (DPs) possess diverse physiological activities. Therefore, in recent years, increasing attention has been drawn to the use of enzymes for the preparation of alginate oligosaccharides for food and industrial applications. Previously, we identified and characterized a novel bifunctional alginate lyase Aly7A, which can specifically release trisaccharide from three different substrate types with a unique degradation pattern. Herein, we investigated its degradation pattern by modular truncation and molecular docking. The results suggested that Aly7A adopted a unique action mode towards different substrates with the substrate chain sliding into the binding pocket of the catalytic domain to position the next trisaccharide for cleavage. Deletion of the Aly7A carbohydrate binding module (CBM) domain resulted in a complex distribution of degradation products and no preference for trisaccharide formation, indicating that the CBM may act as a "controller" during the trisaccharide release process. This study further testifies CBM as a regulator of product distribution and provides new insights into well-defined generation of alginate oligosaccharides with associated CBMs.

Sequence-based prediction of physicochemical interactions at protein functional sites using a function-and-interaction-annotated domain profile database.

BACKGROUND: Identifying protein functional sites (PFSs) and, particularly, the physicochemical interactions at these sites is critical to understanding protein functions and the biochemical reactions involved. Several knowledge-based methods have been developed for the prediction of PFSs; however, accurate methods for predicting the physicochemical interactions associated with PFSs are still lacking. RESULTS: In this paper, we present a sequence-based method for the prediction of physicochemical interactions at PFSs. The method is based on a functional site and physicochemical interaction-annotated domain profile database, called fiDPD, which was built using protein domains found in the Protein Data Bank. This method was applied to 13 target proteins from the very recent Critical Assessment of Structure Prediction (CASP10/11), and our calculations gave a Matthews correlation coefficient (MCC) value of 0.66 for PFS prediction and an 80% recall in the prediction of the associated physicochemical interactions. CONCLUSIONS: Our results show that, in addition to the PFSs, the physical interactions at these sites are also conserved in the evolution of proteins. This work provides a valuable sequence-based tool for rational drug design and side-effect assessment. The method is freely available and can be accessed at http://202.119.249.49 .

Nonnative Energetic Frustrations in Protein Folding at Residual Level: A Simulation Study of Homologous Immunoglobulin-like ß-Sandwich Proteins.

Nonnative interactions cause energetic frustrations in protein folding and were found to dominate key events in folding intermediates. However, systematically characterizing energetic frustrations that are caused by nonnative intra-residue interactions at residual resolution is still lacking. Recently, we studied the folding of a set of homologous all-α proteins and found that nonnative-contact-based energetic frustrations are highly correlated to topology of the protein native-contact network. Here, we studied the folding of nine homologous immunoglobulin-like (Ig-like) ß-sandwich proteins, and examined nonnative-contact-based energetic frustrations Go-like model. Our calculations showed that nonnative-interaction-based energetic frustrations in ß-sandwich proteins are much more complicated than those in all- α proteins, and they exhibit highly heterogeneous effects on the folding of secondary structures. Further, the nonnative interactions introduced distinct correlations in the folding of different folding-patches of ß-sandwich proteins. Taken together, a strong interplay might exist between nonnative-interaction energetic frustrations and the protein native-contact networks, which ensures that ß-sandwich domains adopt a common folding mechanism.

Amino-Acid Network Clique Analysis of Protein Mutation Non-Additive Effects: A Case Study of Lysozme.

Optimizing amino-acid mutations in enzyme design has been a very challenging task in modern bio-industrial applications. It is well known that many successful designs often hinge on extensive correlations among mutations at different sites within the enzyme, however, the underpinning mechanism for these correlations is far from clear. Here, we present a topology-based model to quantitively characterize non-additive effects between mutations. The method is based on the molecular dynamic simulations and the amino-acid network clique analysis. It examines if the two mutation sites of a double-site mutation fall into to a 3-clique structure, and associates such topological property of mutational site spatial distribution with mutation additivity features. We analyzed 13 dual mutations of T4 phage lysozyme and found that the clique-based model successfully distinguishes highly correlated or non-additive double-site mutations from those additive ones whose component mutations have less correlation. We also applied the model to protein Eglin c whose structural topology is significantly different from that of T4 phage lysozyme, and found that the model can, to some extension, still identify non-additive mutations from additive ones. Our calculations showed that mutation non-additive effects may heavily depend on a structural topology relationship between mutation sites, which can be quantitatively determined using amino-acid network k-cliques. We also showed that double-site mutation correlations can be significantly altered by exerting a third mutation, indicating that more detailed physicochemical interactions should be considered along with the network clique-based model for better understanding of this elusive mutation-correlation principle.