Integrating Structural Information to Study the Dynamics of Protein-Protein Interactions in Cells

Contextual AI models for single-cell protein biology

Understanding protein function and developing molecular therapies require deciphering the cell types in which proteins act as well as the interactions between proteins. However, modeling protein interactions across biological contexts remains challenging for existing algorithms. Here, we introduce P innacle , a geometric deep learning approach that generates context-aware protein representations. Leveraging a multi-organ single-cell atlas, PINNACLE learns on contextualized protein interaction networks to produce 394,760 protein representations from 156 cell type contexts across 24 tissues. P innacle 's embedding space reflects cellular and tissue organization, enabling zero-shot retrieval of the tissue hierarchy. Pretrained protein representations can be adapted for downstream tasks: enhancing 3D structure-based representations for resolving immuno-oncological protein interactions, and investigating drugs' effects across cell types. P innacle outperforms state-of-the-art models in nominating therapeutic targets for rheumatoid arthritis and inflammatory bowel diseases, and pinpoints cell type contexts with higher predictive capability than context-free models. P innacle 's ability to adjust its outputs based on the context in which it operates paves way for diverse large-scale context-specific predictions in biology.

How does the same ligand activate signaling of different receptors in TNFR superfamily: a computational study

TNFα is a highly pleiotropic cytokine inducing inflammatory signaling pathways. It is initially presented on plasma membrane of cells (mTNFα), and also exists in a soluble variant (sTNFα) after cleavage. The ligand is shared by two structurally similar receptors, TNFR1 and TNFR2. Interestingly, while sTNFα preferentially stimulates TNFR1, TNFR2 signaling can only be activated by mTNFα. How can two similar receptors respond to the same ligand in such a different way? We employed computational simulations in multiple scales to address this question. We found that both mTNFα and sTNFα can trigger the clustering of TNFR1. The size of clusters induced by sTNFα is constantly larger than the clusters induced by mTNFα. The systems of TNFR2, on the other hand, show very different behaviors. Only when the interactions between TNFR2 are very weak, mTNFα can trigger the receptors to form very large clusters. Given the same weak binding affinity, only small oligomers were obtained in the system of sTNFα. Considering that TNF-mediated signaling is modulated by the ligand-induced clustering of receptors on cell surface, our study provided the mechanistic foundation to the phenomenon that different isoforms of the ligand can lead to highly distinctive signaling patterns for its receptors.

Combining hypothesis- and data-driven neuroscience modeling in FAIR workflows

Modeling in neuroscience occurs at the intersection of different points of view and approaches. Typically, hypothesis-driven modeling brings a question into focus so that a model is constructed to investigate a specific hypothesis about how the system works or why certain phenomena are observed. Data-driven modeling, on the other hand, follows a more unbiased approach, with model construction informed by the computationally intensive use of data. At the same time, researchers employ models at different biological scales and at different levels of abstraction. Combining these models while validating them against experimental data increases understanding of the multiscale brain. However, a lack of interoperability, transparency, and reusability of both models and the workflows used to construct them creates barriers for the integration of models representing different biological scales and built using different modeling philosophies. We argue that the same imperatives that drive resources and policy for data - such as the FAIR (Findable, Accessible, Interoperable, Reusable) principles - also support the integration of different modeling approaches. The FAIR principles require that data be shared in formats that are Findable, Accessible, Interoperable, and Reusable. Applying these principles to models and modeling workflows, as well as the data used to constrain and validate them, would allow researchers to find, reuse, question, validate, and extend published models, regardless of whether they are implemented phenomenologically or mechanistically, as a few equations or as a multiscale, hierarchical system. To illustrate these ideas, we use a classical synaptic plasticity model, the Bienenstock-Cooper-Munro rule, as an example due to its long history, different levels of abstraction, and implementation at many scales.

Understanding the functional role of membrane confinements in TNF-mediated signaling by multiscale simulations

The interaction between TNFα and TNFR1 is essential in maintaining tissue development and immune responses. While TNFR1 is a cell surface receptor, TNFα exists in both soluble and membrane-bound forms. Interestingly, it was found that the activation of TNFR1-mediated signaling pathways is preferentially through the soluble form of TNFα, which can also induce the clustering of TNFR1 on plasma membrane of living cells. We developed a multiscale simulation framework to compare receptor clustering induced by soluble and membrane-bound ligands. Comparing with the freely diffusive soluble ligands, we hypothesize that the conformational dynamics of membrane-bound ligands are restricted, which affects the clustering of ligand-receptor complexes at cell-cell interfaces. Our simulation revealed that only small clusters can form if TNFα is bound on cell surface. In contrast, the clustering triggered by soluble TNFα is more dynamic, and the size of clusters is statistically larger. We therefore demonstrated the impact of membrane-bound ligand on dynamics of receptor clustering. Moreover, considering that larger TNFα-TNFR1 clusters is more likely to provide spatial platform for downstream signaling pathway, our studies offer new mechanistic insights about why the activation of TNFR1-mediated signaling pathways is not preferred by membrane-bound form of TNFα.

A multiscale study on the mechanisms of spatial organization in ligand-receptor interactions on cell surfaces

The binding of cell surface receptors with extracellular ligands triggers distinctive signaling pathways, leading into the corresponding phenotypic variation of cells. It has been found that in many systems, these ligand-receptor complexes can further oligomerize into higher-order structures. This ligand-induced oligomerization of receptors on cell surfaces plays an important role in regulating the functions of cell signaling. The underlying mechanism, however, is not well understood. One typical example is proteins that belong to the tumor necrosis factor receptor (TNFR) superfamily. Using a generic multiscale simulation platform that spans from atomic to subcellular levels, we compared the detailed physical process of ligand-receptor oligomerization for two specific members in the TNFR superfamily: the complex formed between ligand TNFα and receptor TNFR1 versus the complex formed between ligand TNFβ and receptor TNFR2. Interestingly, although these two systems share high similarity on the tertiary and quaternary structural levels, our results indicate that their oligomers are formed with very different dynamic properties and spatial patterns. We demonstrated that the changes of receptor’s conformational fluctuations due to the membrane confinements are closely related to such difference. Consistent to previous experiments, our simulations also showed that TNFR can preassemble into dimers prior to ligand binding, while the introduction of TNF ligands induced higher-order oligomerization due to a multivalent effect. This study, therefore, provides the molecular basis to TNFR oligomerization and reveals new insights to TNFR-mediated signal transduction. Moreover, our multiscale simulation framework serves as a prototype that paves the way to study higher-order assembly of cell surface receptors in many other bio-systems.

A computational study of co-inhibitory immune complex assembly at the interface between T cells and antigen presenting cells

The activation and differentiation of T-cells are mainly directly by their co-regulatory receptors. T lymphocyte-associated protein-4 (CTLA-4) and programed cell death-1 (PD-1) are two of the most important co-regulatory receptors. Binding of PD-1 and CTLA-4 with their corresponding ligands programed cell death-ligand 1 (PD-L1) and B7 on the antigen presenting cells (APC) activates two central co-inhibitory signaling pathways to suppress T cell functions. Interestingly, recent experiments have identified a new cis-interaction between PD-L1 and B7, suggesting that a crosstalk exists between two co-inhibitory receptors and the two pairs of ligand-receptor complexes can undergo dynamic oligomerization. Inspired by these experimental evidences, we developed a coarse-grained model to characterize the assembling of an immune complex consisting of CLTA-4, B7, PD-L1 and PD-1. These four proteins and their interactions form a small network motif. The temporal dynamics and spatial pattern formation of this network was simulated by a diffusion-reaction algorithm. Our simulation method incorporates the membrane confinement of cell surface proteins and geometric arrangement of different binding interfaces between these proteins. A wide range of binding constants was tested for the interactions involved in the network. Interestingly, we show that the CTLA-4/B7 ligand-receptor complexes can first form linear oligomers, while these oligomers further align together into two-dimensional clusters. Similar phenomenon has also been observed in other systems of cell surface proteins. Our test results further indicate that both co-inhibitory signaling pathways activated by B7 and PD-L1 can be down-regulated by the new cis-interaction between these two ligands, consistent with previous experimental evidences. Finally, the simulations also suggest that the dynamic and the spatial properties of the immune complex assembly are highly determined by the energetics of molecular interactions in the network. Our study, therefore, brings new insights to the co-regulatory mechanisms of T cell activation.

The activation of a T cell can be regulated by the receptors on its surface, such as CTLA-4 and PD-1. People used to think that these two receptors inhibit T cell activation through distinct pathways. However, recent experiments discovered that the ligands of these two receptors, B7 and PD-L1, can interact with each other on the same surface of antigen presenting cells. Here we utilized computational simulations to investigate functional roles of this newly discovered interaction in T cell coregulation. The specific environment of interface between T cell and antigen presenting cell has been taken into account of our model. Ligand and receptors randomly diffuse within this interface area. They further involve in different types of interactions, with each other from the same side or the opposite side of cell surface. Using this method, we found ligands and receptors can not only form complexes, but also aggregate into large-scale clusters. We also demonstrated that the engagement between B7 and PD-L1 can reduce the interactions with their corresponding receptors. This study, therefore, offers new insights to our understanding of signal regulation in T cells.

Characterizing the function of domain linkers in regulating the dynamics of multi-domain fusion proteins by microsecond molecular dynamics simulations and artificial intelligence

Multi-domain proteins are not only formed through natural evolution but can also be generated by recombinant DNA technology. Because many fusion proteins can enhance the selectivity of cell targeting, these artificially produced molecules, called multi-specific biologics, are promising drug candidates, especially for immunotherapy. Moreover, the rational design of domain linkers in fusion proteins is becoming an essential step toward a quantitative understanding of the dynamics in these biopharmaceutics. We developed a computational framework to characterize the impacts of peptide linkers on the dynamics of multi-specific biologics. Specifically, we first constructed a benchmark containing six types of linkers that represent various lengths and degrees of flexibility and used them to connect two natural proteins as a test system. We then projected the microsecond dynamics of these proteins generated from Anton onto a coarse-grained conformational space. We further analyzed the similarity of dynamics among different proteins in this low-dimensional space by a neural-network-based classification model. Finally, we applied hierarchical clustering to place linkers into different subgroups based on the classification results. The clustering results suggest that the length of linkers, which is used to spatially separate different functional modules, plays the most important role in regulating the dynamics of this fusion protein. Given the same number of amino acids, linker flexibility functions as a regulator of protein dynamics. In summary, we illustrated that a new computational strategy can be used to study the dynamics of multi-domain fusion proteins by a combination of long timescale molecular dynamics simulation, coarse-grained feature extraction, and artificial intelligence.

Understanding the impacts of cellular environments on ligand binding of membrane receptors by computational simulations

Binding of cell surface receptors with their extracellular ligands initiates various intracellular signaling pathways. However, our understanding of the cellular functions of these receptors is very limited due to the fact that in vivo binding between ligands and receptors has only been successfully measured in a very small number of cases. In living cells, receptors are anchored on surfaces of the plasma membrane, which undergoes thermal undulations. Moreover, it has been observed in various systems that receptors can be organized into oligomers prior to ligand binding. It is not well understood how these cellular factors play roles in regulating the dynamics of ligand-receptor interactions. Here, we tackled these problems by using a coarse-grained kinetic Monte Carlo simulation method. Using this method, we demonstrated that the membrane undulations cause a negative effect on ligand-receptor interactions. We further found that the preassembly of membrane receptors on the cell surface can not only accelerate the kinetics of ligand binding but also reduce the noises during the process. In general, our study highlights the importance of membrane environments in regulating the function of membrane receptors in cells. The simulation method can be potentially applied to specific receptor systems involved in cell signaling.

Understand the Functions of Scaffold Proteins in Cell Signaling by a Mesoscopic Simulation Method

Scaffold proteins are central players in regulating the spatial-temporal organization of many important signaling pathways in cells. They offer physical platforms to downstream signaling proteins so that their transient interactions in a crowded and heterogeneous environment of cytosol can be greatly facilitated. However, most scaffold proteins tend to simultaneously bind more than one signaling molecule, which leads to the spatial assembly of multimeric protein complexes. The kinetics of these protein oligomerizations are difficult to quantify by traditional experimental approaches. To understand the functions of scaffold proteins in cell signaling, we developed a, to our knowledge, new hybrid simulation algorithm in which both spatial organization and binding kinetics of proteins were implemented. We applied this new technique to a simple network system that contains three molecules. One molecule in the network is a scaffold protein, whereas the other two are its binding targets in the downstream signaling pathway. Each of the three molecules in the system contains two binding motifs that can interact with each other and are connected by a flexible linker. By applying the new simulation method to the model, we show that the scaffold proteins will promote not only thermodynamics but also kinetics of cell signaling given the premise that the interaction between the two signaling molecules is transient. Moreover, by changing the flexibility of the linker between two binding motifs, our results suggest that the conformational fluctuations in a scaffold protein play a positive role in recruiting downstream signaling molecules. In summary, this study showcases the capability of computational simulation in understanding the general principles of scaffold protein functions.

Cadherin clusters stabilized by a combination of specific and nonspecific cis-interactions

We demonstrate a combined experimental and computational approach for the quantitative characterization of lateral interactions between membrane-associated proteins. In particular, weak, lateral (cis) interactions between E-cadherin extracellular domains tethered to supported lipid bilayers, were studied using a combination of dynamic single-molecule Förster Resonance Energy Transfer (FRET) and kinetic Monte Carlo (kMC) simulations. Cadherins are intercellular adhesion proteins that assemble into clusters at cell-cell contacts through cis- and trans- (adhesive) interactions. A detailed and quantitative understanding of cis-clustering has been hindered by a lack of experimental approaches capable of detecting and quantifying lateral interactions between proteins on membranes. Here single-molecule intermolecular FRET measurements of wild-type E-cadherin and cis-interaction mutants combined with simulations demonstrate that both nonspecific and specific cis-interactions contribute to lateral clustering on lipid bilayers. Moreover, the intermolecular binding and dissociation rate constants are quantitatively and independently determined, demonstrating an approach that is generalizable for other interacting proteins.

Understanding the Impacts of Conformational Dynamics on the Regulation of Protein-Protein Association by a Multiscale Simulation Method

Complexes formed among diverse proteins carry out versatile functions in nearly all physiological processes. Association rates which measure how fast proteins form various complexes are of fundamental importance to characterize their functions. The association rates are not only determined by the energetic features at binding interfaces of a protein complex but also influenced by the intrinsic conformational dynamics of each protein in the complex. Unfortunately, how this conformational effect regulates protein association has never been calibrated on a systematic level. To tackle this problem, we developed a multiscale strategy to incorporate the information on protein conformational variations from Langevin dynamic simulations into a kinetic Monte Carlo algorithm of protein-protein association. By systematically testing this approach against a large-scale benchmark set, we found the association of a protein complex with a relatively rigid structure tends to be reduced by its conformational fluctuations. With specific examples, we further show that higher degrees of structural flexibility in various protein complexes can facilitate the searching and formation of intermolecular interactions and thereby accelerate their associations. In general, the integration of conformational dynamics can improve the correlation between experimentally measured association rates and computationally derived association probabilities. Finally, we analyzed the statistical distributions of different secondary structural types on protein-protein binding interfaces and their preference to the change of association rates. Our study, to the best of our knowledge, is the first computational method that systematically estimates the impacts of protein conformational dynamics on protein-protein association. It throws lights on the molecular mechanisms of how protein-protein recognition is kinetically modulated.

Using Coarse-Grained Simulations to Characterize the Mechanisms of Protein-Protein Association

The formation of functionally versatile protein complexes underlies almost every biological process. The estimation of how fast these complexes can be formed has broad implications for unravelling the mechanism of biomolecular recognition. This kinetic property is traditionally quantified by association rates, which can be measured through various experimental techniques. To complement these time-consuming and labor-intensive approaches, we developed a coarse-grained simulation approach to study the physical processes of protein–protein association. We systematically calibrated our simulation method against a large-scale benchmark set. By combining a physics-based force field with a statistically-derived potential in the simulation, we found that the association rates of more than 80% of protein complexes can be correctly predicted within one order of magnitude relative to their experimental measurements. We further showed that a mixture of force fields derived from complementary sources was able to describe the process of protein–protein association with mechanistic details. For instance, we show that association of a protein complex contains multiple steps in which proteins continuously search their local binding orientations and form non-native-like intermediates through repeated dissociation and re-association. Moreover, with an ensemble of loosely bound encounter complexes observed around their native conformation, we suggest that the transition states of protein–protein association could be highly diverse on the structural level. Our study also supports the idea in which the association of a protein complex is driven by a “funnel-like” energy landscape. In summary, these results shed light on our understanding of how protein–protein recognition is kinetically modulated, and our coarse-grained simulation approach can serve as a useful addition to the existing experimental approaches that measure protein–protein association rates.

A Systematic Test of Receptor Binding Kinetics for Ligands in Tumor Necrosis Factor Superfamily by Computational Simulations

Ligands in the tumor necrosis factor (TNF) superfamily are one major class of cytokines that bind to their corresponding receptors in the tumor necrosis factor receptor (TNFR) superfamily and initiate multiple intracellular signaling pathways during inflammation, tissue homeostasis, and cell differentiation. Mutations in the genes that encode TNF ligands or TNFR receptors result in a large variety of diseases. The development of therapeutic treatment for these diseases can be greatly benefitted from the knowledge on binding properties of these ligand–receptor interactions. In order to complement the limitations in the current experimental methods that measure the binding constants of TNF/TNFR interactions, we developed a new simulation strategy to computationally estimate the association and dissociation between a ligand and its receptor. We systematically tested this strategy to a comprehensive dataset that contained structures of diverse complexes between TNF ligands and their corresponding receptors in the TNFR superfamily. We demonstrated that the binding stabilities inferred from our simulation results were compatible with existing experimental data. We further compared the binding kinetics of different TNF/TNFR systems, and explored their potential functional implication. We suggest that the transient binding between ligands and cell surface receptors leads into a dynamic nature of cross-membrane signal transduction, whereas the slow but strong binding of these ligands to the soluble decoy receptors is naturally designed to fulfill their functions as inhibitors of signal activation. Therefore, our computational approach serves as a useful addition to current experimental techniques for the quantitatively comparison of interactions across different members in the TNF and TNFR superfamily. It also provides a mechanistic understanding to the functions of TNF-associated cell signaling pathways.

A computational model for understanding the oligomerization mechanisms of TNF receptor superfamily

By recognizing members in the tumor necrosis factor (TNF) receptor superfamily, TNF ligand proteins function as extracellular cytokines to activate various signaling pathways involved in inflammation, proliferation, and apoptosis. Most ligands in TNF superfamily are trimeric and can simultaneously bind to three receptors on cell surfaces. It has been experimentally observed that the formation of these molecular complexes further triggers the oligomerization of TNF receptors, which in turn regulate the intracellular signaling processes by providing transient compartmentalization in the membrane proximal regions of cytoplasm. In order to decode the molecular mechanisms of oligomerization in TNF receptor superfamily, we developed a new computational method that can physically simulate the spatial-temporal process of binding between TNF ligands and their receptors. The simulations show that the TNF receptors can be organized into hexagonal oligomers. The formation of this spatial pattern is highly dependent not only on the molecular properties such as the affinities of trans and cis binding, but also on the cellular factors such as the concentration of TNF ligands in the extracellular area or the density of TNF receptors on cell surfaces. Moreover, our model suggests that if TNF receptors are pre-organized into dimers before ligand binding, these lateral interactions between receptor monomers can play a positive role in stabilizing the ligand-receptor interactions, as well as in regulating the kinetics of receptor oligomerization. Altogether, this method throws lights on the mechanisms of TNF ligand-receptor interactions in cellular environments.

Multiscale simulation unravel the kinetic mechanisms of inflammasome assembly

In the innate immune system, the host defense from the invasion of external pathogens triggers the inflammatory responses. Proteins involved in the inflammatory pathways were often found to aggregate into supramolecular oligomers, called 'inflammasome', mostly through the homotypic interaction between their domains that belong to the death domain superfamily. Although much has been known about the formation of these helical molecular machineries, the detailed correlation between the dynamics of their assembly and the structure of each domain is still not well understood. Using the filament formed by the PYD domains of adaptor molecule ASC as a test system, we constructed a new multiscale simulation framework to study the kinetics of inflammasome assembly. We found that the filament assembly is a multi-step, but highly cooperative process. Moreover, there are three types of binding interfaces between domain subunits in the ASC^PYD filament. The multiscale simulation results suggest that dynamics of domain assembly are rooted in the primary protein sequence which defines the energetics of molecular recognition through three binding interfaces. Interface I plays a more regulatory role than the other two in mediating both the kinetics and the thermodynamics of assembly. Finally, the efficiency of our computational framework allows us to design mutants on a systematic scale and predict their impacts on filament assembly. In summary, this is, to the best of our knowledge, the first simulation method to model the spatial-temporal process of inflammasome assembly. Our work is a useful addition to a suite of existing experimental techniques to study the functions of inflammasome in innate immune system.

Sequence-Defined Macrocycles for Understanding and Controlling the Build-up of Hierarchical Order in Self-Assembled 2D Arrays

Anfinsen's dogma that sequence dictates structure is fundamental to understanding the activity and assembly of proteins. This idea has been applied to all manner of oligomers but not to the behavior of cyclic oligomers, aka macrocycles. We do this here by providing the first proofs that sequence controls the hierarchical assembly of nonbiological macrocycles, in this case, at graphite surfaces. To design macrocycles with one (AAA), two (AAB), or three (ABC) different carbazole units, we needed to subvert the synthetic preferences for one-pot macrocyclizations. We developed a new stepwise synthesis with sequence-defined targets made in 11, 17, and 22 steps with 25, 10, and 5% yields, respectively. The linear build up of primary sequence (1°) also enabled a thermal Huisgen cycloaddition to proceed regioselectively for the first time using geometric control. The resulting macrocycles are planar (2° structure) and form H-bonded dimers (3°) at surfaces. Primary sequences encoded into the suite of tricarb macrocycles were shown by scanning-tunneling microscopy (STM) to impact the next levels of supramolecular ordering (4°) and 2D crystalline polymorphs (5°) at solution-graphite interfaces. STM imaging of an AAB macrocycle revealed the formation of a new gap phase that was inaccessible using only C3-symmetric macrocycles. STM imaging of two additional sequence-controlled macrocycles (AAD, ABE) allowed us to identify the factors driving the formation of this new polymorph. This demonstration of how sequence controls the hierarchical patterning of macrocycles raises the importance of stepwise syntheses relative to one-pot macrocyclizations to offer new approaches for greater understanding and control of hierarchical assembly.

A Multiscale Model for the Self-Assembly of Coat Proteins in Bacteriophage MS2

The self-assembly of viral capsids is an essential step to the formation of infectious viruses. Elucidating the kinetic mechanisms of how a capsid or virus-like particle assembles could advance our knowledge about the viral lifecycle, as well as the general principles in self-assembly of biomaterials. However, current understanding of capsid assembly remains incomplete for many viruses due to the fact that the transient intermediates along the assembling pathways are experimentally difficult to be detected. In this paper, we constructed a new multiscale computational framework to simulate the self-assembly of virus-like particles. We applied our method to the coat proteins of bacteriophage MS2 as a specific model system. This virus-like particle of bacteriophage MS2 has a unique feature that its 90 sequence-identical dimers can be classified into two structurally various groups: one is the symmetric CC dimer, and the other is the asymmetric AB dimer. The homotypic interactions between AB dimers result in a 5-fold symmetric contact, while the heterotypic interactions between AB and CC dimers result in 6-fold symmetric contact. We found that the assembly can be described as a physical process of phase transition that is regulated by various factors such as concentration and specific stoichiometry between AB and CC dimers. Our simulations also demonstrate that heterotypic and homotypic interfaces play distinctive roles in modulating the assembling kinetics. The interaction between AB and CC dimers is much more dynamic than that between two AB dimers. We therefore suggest that the alternate growth of viral capsid through the heterotypic dimer interactions dominates the assembling pathways. This is, to the best of our knowledge, the first multiscale model to simulate the assembling process of coat proteins in bacteriophage MS2. The generality of this approach opens the door to its further applications in assembly of other viral capsids, virus-like particles, and novel drug delivery systems.

Computational studies of protein-protein dissociation by statistical potential and coarse-grained simulations: a case study on interactions between colicin E9 endonuclease and immunity proteins

Proteins carry out their diverse functions in cells by forming interactions with each other. The dynamics of these interactions are quantified by the measurement of association and dissociation rate constants. Relative to the efforts made to model the association of biomolecules, little has been studied to understand the principles of protein complex dissociation. Using the interaction between colicin E9 endonucleases and immunity proteins as a test system, here we develop a coarse-grained simulation method to explore the dissociation mechanisms of protein complexes. The interactions between proteins in the complex are described by the knowledge-based potential that was constructed by the statistics from available protein complexes in the structural database. Our study provides the supportive evidences to the dual recognition mechanism for the specificity of binding between E9 DNase and immunity proteins, in which the conserved residues of helix III of Im2 and Im9 proteins act as the anchor for binding, while the sequence variations in helix II make positive or negative contributions to specificity. Beyond that, we further suggest that this binding specificity is rooted in the process of complex dissociation instead of association. While we increased the flexibility of protein complexes, we further found that they are less prone to dissociation, suggesting that conformational fluctuations of protein complexes play important functional roles in regulating their binding and dissociation. Our studies therefore bring new insights to the molecule mechanisms of protein-protein interactions, while the method can serve as a new addition to a suite of existing computational tools for the simulations of protein complexes.