Research Overview
The current research is focussed on the understanding, prediction, and modulation of interactions involving biological macromolecules. We apply and develop techniques grounded in bioinformatics, computational biophysics, and computational pharmaceutical chemistry such as docking, molecular dynamics simulations, free energy estimations, and methods to characterize the flexibility of protein and RNA structures. Three main research areas can be distinguished:
I. Structure, function, and molecular recognition investigated by molecular dynamics simulations
|
||
|
Antibiotics binding and protein voyage in the ribosomal exit tunnel
Ribosomes are large ribonucleoprotein complexes and the protein synthesis machines of the cell. After peptide bond formation, the nascent polypeptides leave the peptidyl transferase center via the ribosomal exit tunnel (RET), which spans the entire large ribosomal subunit. The ribosome is also one of the major targets within the cell for antimicrobials, many of which execute their inhibitory effect by binding at the PTC or in the RET. We use molecular dynamics (MD) simulations to obtain insights in atomic detail into RET dynamics and the protein voyage, and combine MD with free energy calculations [16,18] to investigate antibiotics binding. Furthermore, to understand how RNA molecules undergo large but controlled conformational changes to achieve important functional roles, concepts of graph theory and rigidity theory are applied to a topological network representation of the ribosome to investigate static properties of the RET [33,40,41,45]. |
The large ribosomal subunit. |
|
|
Structure of a riboswitch bound to a ligand. |
Conformational heterogeneity of RNA and proteins Small RNAs, such as the HIV-1 Trans Activation Response (TAR) element and riboswitches, as well as proteins, such as α5β1-integrin, are subject to conformational changes upon binding of ligands or changes in their environment. Correctly describing and predicting such conformational changes is key for linking the biomolecules' structure and function. In the case of TAR RNA, MD and constrained geometrical simulations are performed on the unbound structure in order to test whether bound conformations can be generated that can be successfully used for docking [51]. For α5β1-integrin, recent experimental findings suggested an activation by high urea concentrations. MD simulations at such concentrations indeed showed conformational changes that resulted in an unbending of the integrin structure, which is necessary for the activation [52]. At present, structural changes in integrin upon binding of agonists and anagonists are investigated by MD simulations. In order to characterize the unbound state of riboswitches, we currently use MD simulations to gain deeper insights into the riboswitch's conformational heterogeneity and binding pathways. |
|
|
Binding free energy and entropy calculations
Understanding the binding properties of ligands, proteins, and nucleic acids is essential for designing new molecules with good binding properties. Based on conformational ensembles from molecular dynamics simulations, we determine the binding affinity of complex components by implicit solvent free energy calculations, thermodynamic integration, and entropy evaluation methods. An analysis of the binding free energy of self-paired fluorinated benzene bases that are known to increasingly stabilize RNA duplexes with increasing number of fluorine atoms provided detailed insights into the determinants of stability differences on a atomic level [36]. In a performance study focussing on lead optimization, we aim at identifying procedures and strategies that allow to correctly rank potential drug candidates according to their binding affinity. In this context, correctly and efficiently estimating translational, rotational, and configurational entropy contributions to the binding free energy is still a holy grail [18]. We aim at developing such a procedure that can be used as a pivotal tool in computational lead optimization techniques. |
Binding free energy contributions in Factor Xa. |
|
|
|
||
|
Protein-ligand and protein-protein interactions play an important role for the function of living organisms. Thus, one of the research areas of our group is focussed on analyzing regions where these interactions occur. In particular, we focus on aspects of plasticity and "ligandability" of protein-protein interfaces and ligand binding sites and protein function prediction by binding pocket comparison. |
||
|
Structural and energetic determinants of protein-protein binding At present, the design of small-molecule protein-protein interactions modulators (PPIM) encounters at least two challenges [25,64]. First, protein-protein interfaces are rather flat and usually lack a distinct binding pocket. Second, due to the size of protein-protein interfaces interactions that are favorable for binding can be widely distributed. We address these challenges based on the protein-protein complex structure. Therefore, we sample accessible conformations by molecular dynamics (MD) simulation [21] and the fast geometrical simulation method FRODA [55,59,60], predict the opening of transient binding pockets in the interface, identify "hot spot" residues that participate in important interactions, and eventually design peptidomimetic or screen for small-molecule PPIM.
|
Ligand binding to a transient pocket in IL-2.
|
|
|
Knowledge-driven design of protein-protein interaction modulators. |
Knowledge-driven modulation of protein-protein interactions The overall strategy, exclusively based on the protein-protein structure, comprises four steps. (I) Hot spots of protein-protein binding are identified by MM-PB/SA free energy decomposition [16,18,19,48] and knowledge-based DrugScore interactions potentials [4,12,14,23,30,50,56]. (II) These hot spots are validated experimentally by alanine mutations. (III) Selected hot spots are used as a query for similarity screening of small-molecule databases and the design of peptidomimetic compounds. Subsequently, potential PPIM are investigated by molecular docking and ranked by free energy calculations. (IV) Predicted PPIM are tested experimentally. We currently focus on important pharmacological targets such as IL‑2/IL‑2Rα, the HSP90 dimer, and NHR2 [48]. |
|
|
Characterizing protein binding sites to predict druggability and protein function
Since the shape, volume, and size of binding pockets usually varies, up to now there is no standard definition what constitutes a binding pocket, and descriptors allowing an in-depth characterization of binding pockets are lacking. Thus, we aim at developing novel approaches to characterize protein binding pockets based on knowledge-based potential fields [4]. The resulting binding site descriptors can be used to predict target ligandability, i.e. the ability of a binding pocket to bind small molecules with high affinity. Additionally, this characterization of binding pockets allows protein function de-orphanization by comparing binding pockets of multiple proteins and transferring the function of the most similar protein to the protein with unknown function. Additionally, we maintain an inhouse binding interface (BIF) database that allows a large-scale exploration of protein-protein interactions that putatively can be inhibited by helix mimetics. |
HIV-1 protease binding site in complex with darunavir and |
|
|
|
||
III. Modeling flexibility and plasticity of biomolecules / Development of novel docking approachesModeling protein flexibility and plasticity is important for understanding the function of biological systems. Given a network representation of a macromelcular structure, its intrinsic flexibility can be efficiently identified by determining the number and spatial distribution of bond-rotational degrees of freedom in the network. The resulting rigid cluster decomposition can be further used as input of coarse graininig schemes such as normal mode analysis to model biomolecular mobility. This information can then be incorporated into docking algorithms to efficiently consider receptor flexibility. |
||
|
Protein rigidity, thermal adaptation, and allostery
Biomacromolecules require a balance of flexibility and rigidity to achieve their diverse functional roles. Hence, it is desirable to have a precise knowledge about what can move and how. The flexibility for a given biomacromolecule can be analyzed within a few seconds by a graph theory-based approach as implemented in the FIRST program by Thorpe et al. For further quantifying biomacromolecular stability, the Constraint Network Analysis (CNA) program package is being developed in our group. This program functions as a front-end to FIRST and allows to I) set up a variety of constraint network representations for rigidity analysis, II) process the results obtained from FIRST, and III) calculate different indices for characterizing the macroscopic and microscopic stability in biomacromolecules. FIRST and CNA were successfully applied to analyze changes in protein flexibility upon complex formation [19], to investigate the stability of proteins from mesophilic and thermophilic organisms [37,55], and to characterize the flexibility of RNA structures [33,41] and the ribosomal exit tunnel [40]. Currently, the CNA program package is used for predicting the coupling between allosteric sites by identifying physically connected pathways of residues in proteins. |
Rigid cluster decomposition and stability map from meso- |
|
Example of an elastic grid-based protein-docking based on |
Fully-flexible docking based on an elastic grid representation of potential fields Most current docking approaches consider the receptor as rigid. However, this simplification can lead to the failure of docking attempts in those cases where proteins or nucleic acids undergo conformational changes during complex formation [28,35,51]. Thus, we develop a fully-flexible docking approach that uses an elastic grid-based representation of potential fields as objective function. The elastic grid thereby adapts to movements occuring in one or both of the binding partners. Several knowledge-based scoring functions have been developed in the group for describing protein-protein [50,56], protein-ligand [4] and protein-RNA [30] interactions. In total, we expect to get to a docking approach that yields accurate results in cases of conformational variability of the macromolecules, yet is still efficient [42,57]. |
|
|
Multi-scale modeling of macromolecular conformational transitions Modeling conformational transitions of macromolecules is computationally challenging. Recently, coarse-grained normal mode approaches based on elastic network theory have emerged as efficient alternatives for investigating large-scale conformational changes. It has been shown that functionally important conformational changes of proteins can be described by low frequency normal modes, which are robust and insensitive to higher coarse-graining [53]. Accordingly, we have introduced a three-step approach for multi-scale modeling of macromolecular conformational changes. The first two steps are based on recent developments in rigidity and elastic network theory (termed Rigid Cluster Normal Mode Analysis) [24,26]. In the final step, the recently introduced idea of constrained geometric simulations of diffusive motions in proteins is extended. New macromolecule conformers are generated by deforming the structure along low-energy normal mode directions predicted by RCNMA plus random direction components [58]. |
Three-step approach for multi-scale modeling |
|