NMR-based docking of protein-protein complexes : the human UbcH5b-CNOT4 ubiquitination complex

Understanding the molecular and functional interactions among macromolecular complexes, as well as their changes associated with time, cell type or disease state will be invaluable for human health, and will have direct implications, for example, in pharmaceutical research to identify and select potential targets, and design efficient and specific drugs. Structural studies of macromolecular complexes, however, suffer from some limitations, especially in the case of weak and transient complexes. New and complementary methodologies, such as docking, have therefore been developed. The current computational approaches, however, also suffer from limitations and new developments and improvements are needed. This thesis introduces a new docking approach in which the docking of two macromolecules is driven by biophysical and/or biochemical information and furthermore describes structural studies on the UbcH5B-CNOT4 complex involved in the ubiquitination pathway. HADDOCK is a new docking approach that is based on biophysical and/or biochemical information. This information, derived for example from NMR chemical shift perturbation or site-directed mutagenesis experiments, is converted into highly ambiguous intermolecular distance restraints that are directly used to drive the docking process. The docking protocol allows for side chains and backbone flexibility at the interface and the solutions are scored according to an intermolecular energy term. The method was successfully tested on three complexes. The solution NMR structure of UbcH5B, an E2 ubiquitin conjugating enzyme has been solved. NMR relaxation measurements are performed on UbcH5B. They show limited motions for the major part of the protein backbone. We compare the structure of UbcH5B with other E2 structures, and the global fold of all E2s is very similar. Some differences are, however, observed and correlate well with the dynamical properties of E2s. The position and orientation of the N-terminal a-helix as compared to the core of the protein differ in the various structures. This difference may be determinant in E3 ubiquitin ligase binding and recognition. Furthermore, a highly conserved asparagine residue was shown to be important for the ubiquitin transfer. In crystal structures, this asparagine points away from the active site cysteine. Structure of UbcH5B shows that in solution, this asparagine is in close proximity to the active site cysteine, in a conformation suitable for its catalytic role. HADDOCK is then used to generate a structural model of the UbcH5B-CNOT4 complex. CNOT4 is an E3 ubiquitin ligase that is part of the CCR4-NOT complex involved in transcription repression. The residues of the CNOT4 RING domain important for the interaction with UbcH5B were previously reported. Here, the residues of UbcH5B important for the binding to CNOT4 RING are identified from NMR chemical shift perturbation experiments. These data are used to generate a structural model of the UbcH5B-CNOT4 complex. Two sets of solutions are, however, obtained that can not be discriminated. Mutagenesis experiments are performed and identify charged residues of UbcH5B (Lys63) and of CNOT4 (Glu49) involved in an electrostatic interaction. Once this information is included in the docking, a unique set of solutions is obtained. The structural model of the UbcH5B-CNOT4 complex is compared with the X-ray structure of the homologous UbcH7-c-Cbl complex and significant differences at specific residues give structural insights into the mechanisms of the E2-E3 specificity