Targeting the cryptic sites: NMR-based strategy to improve protein druggability by controlling the conformational equilibrium.

Cryptic ligand binding sites, which are not evident in the unligated structures, are beneficial in tackling with difficult but attractive drug targets, such as protein-protein interactions (PPIs). However, cryptic sites have thus far not been rationally pursued in the early stages of drug development. Here, we demonstrated by nuclear magnetic resonance that the cryptic site in Bcl-xL exists in a conformational equilibrium between the open and closed conformations under the unligated condition. While the fraction of the open conformation in the unligated wild-type Bcl-xL is estimated to be low, F143W mutation that is distal from the ligand binding site can substantially elevate the population. The F143W mutant showed a higher hit rate in a phage-display peptide screening, and the hit peptide bound to the cryptic site of the wild-type Bcl-xL. Therefore, by controlling the conformational equilibrium in the cryptic site, the opportunity to identify a PPI inhibitor could be improved.

Improvement of Ligand Affinity and Thermodynamic Properties by NMR-Based Evaluation of Local Dynamics and Surface Complementarity in the Receptor-Bound State.

The thermodynamic properties of a ligand in the bound state affect its binding specificity. Strict binding specificity can be achieved by introducing multiple spatially defined interactions, such as hydrogen bonds and van der Waals interactions, into the ligand-receptor interface. These introduced interactions are characterized by restricted local dynamics and improved surface complementarity in the bound state. In this study, we experimentally evaluated the local dynamics and the surface complementarity of weak-affinity ligands in the receptor-bound state by forbidden coherence transfer analysis in free-bound exchange systems (Ex-FCT), using the interaction between a ligand, a myocyte-enhancer factor 2A (MEF2A) docking peptide, and a receptor, p38α, as a model system. The Ex-FCT analyses successfully provided information for the rational design of a ligand with higher affinity and preferable thermodynamic properties for p38α.

Suppression of problematic compound oligomerization by cosolubilization of nondetergent sulfobetaines.

Numerous small organic compounds exist in equilibrium among monomers, soluble oligomers, and insoluble aggregates in aqueous solution. Compound aggregation is a major reason for false positives in drug screening, and even soluble oligomers can interfere with structural and biochemical analyses. However, an efficient way to manage the equilibrium of aggregation-prone compounds, especially those involved with soluble oligomers, has not been established. In this study, solution NMR spectroscopy was used as a suitable technique to detect compound oligomers in equilibrium, and it was demonstrated that cosolubilization of nondetergent sulfobetaines (NDSBs) can largely suppress compound oligomerization and aggregation by shifting the equilibrium toward the monomers. The rotational correlation time was obtained from the ratio of the selective and nonselective longitudinal NMR relaxation times, which directly and quantitatively reflected the apparent sizes of the compounds in the equilibrium. The rotational correlation time of the aggregation-prone compound SKF86002 (1 mM) was substantially reduced from 0.31 to 0.23 ns by cosolubilization of 100 mM NDSB195. NDSB cosolubilization allowed us to perform successful structural and biochemical experiments with substantially fewer artifacts, which represents a strategy to directly resolve the problematic oligomerization and aggregation of compounds.

Protein-ligand docking guided by ligand pharmacophore-mapping experiment by NMR.

We developed a new protein-ligand docking calculation method using experimental NMR data. Recently, we proposed a novel ligand epitope-mapping experiment, which utilizes the difference between the longitudinal relaxation rates of ligand protons with and without irradiation of target protein protons (DIRECTION epitope-mapping experiment; Y. Mizukoshi, et al., An accurate pharmacophore mapping method by NMR, submitted for publication). Although the epitope-mapping experiment is simple and rapid, the result should reflect the proximity of ligand protons to the target protein surface. However, it cannot directly provide the protein-ligand complex structure without any other structural information. While the accuracy of protein-ligand docking software is insufficient, the software can provide many candidate complex structures. In many cases, the correct complex structure is included in the set of predicted complex structures and the correct structures could be selected by applying the above experimental result of ligand epitope mapping. In the current study, we combined the protein-ligand docking software with the NMR experimental information so as to improve the prediction of the protein-ligand complex structure. Consequently, the prediction accuracy was improved by 1.3-1.9 times (from ca. 50% to ca. 70%) in a self-docking test for the simulated epitope mapping result. Moreover, this method was applied to actual NMR experiments, and it successfully reconstructed the protein-ligand complex structures.

Precise structural determination of weakly binding peptides by utilizing dihedral angle constraints.

Structural determination of target-bound conformations of peptides is of primary importance for the optimization of peptide ligands and peptide-mimetic design. In the structural determination of weakly binding ligands, transferred nuclear Overhauser effect (TrNOE) methods have been widely used. However, not many distance constraints can be obtained from small peptide ligands by TrNOE, especially for peptides bound to a target molecule in an extended conformation. Therefore, for precise structural determination of weakly binding peptides, additional structural constraints are required. Here, we present a strategy to systematically introduce dihedral angle constraints obtained from multiple transferred cross-correlated relaxation experiments and demonstrate precise structures of weakly binding peptides. As a result, we could determine the bioactive conformations of phage-derived peptide ligands and define their core binding motifs.

Structural and interaction analysis of glycoprotein VI-binding peptide selected from a phage display library.

Glycoprotein VI (GPVI) is a major collagen receptor on the platelet surface that recognizes the glycine-proline-hydroxyproline (GPO) sequence in the collagen molecule and plays a crucial role in thrombus formation. Inhibitors that block the interaction of GPVI with collagen have potential for use as antithrombotic drugs. For low molecular weight drug design for GPVI, it is essential to obtain precise structural and interaction information about GPVI-binding ligands. However, experimentally obtained structural and interaction information of small ligands, such as peptides, in the GPVI-bound state has not been reported. In this study, by screening a phage-displayed peptide library, we discovered a novel peptide ligand (pep-10L; YSDTDWLYFSTS) without any similarities to the sequence of collagen that inhibits GPVI-GPO binding. Systematic Ala scanning in surface plasmon resonance experiments and a saturation transfer difference NMR experiment revealed that Trp(6), Leu(7), Phe(9), and Ser(10) residues in the pep-10L peptide interacted with GPVI. Furthermore, the GPVI-bound conformation of the pep-10L peptide was determined using transferred nuclear Overhauser effect analysis. The obtained structure has revealed that the central part of pep-10L (Asp(5)-Phe(9)) has a helical conformation, the side chains of Trp(6), Leu(7), and Phe(9) form a hydrophobic side in the helix, and the Tyr(8) side chain faces the opposite direction from the hydrophobic side. Computational docking prediction has shown that the hydrophobic side of pep-10L sticks in the hydrophobic groove on the GPVI surface, which corresponds to the putative collagen-related peptide binding groove. These data could enable the structure-guided development of a small molecule GPVI antagonist.

Utilization of methyl proton resonances in cross-saturation measurement for determining the interfaces of large protein-protein complexes.

Cross-saturation experiments allow the identification of the contact residues of large protein complexes (MW>50 K) more rigorously than conventional NMR approaches which involve chemical shift perturbations and hydrogen-deuterium exchange experiments [Takahashi et al. (2000) Nat. Struct. Biol., 7, 220-223]. In the amide proton-based cross-saturation experiment, the combined use of high deuteration levels for non-exchangeable protons of the ligand protein and a solvent with a low concentration of (1)H(2)O greatly enhanced the selectivity of the intermolecular cross-saturation phenomenon. Unfortunately, experimental limitations caused losses in sensitivity. Furthermore, since main chain amide protons are not generally exposed to solvent, the efficiency of the saturation transfer directed to the main chain amide protons is not very high. Here we propose an alternative cross-saturation experiment which utilizes the methyl protons of the side chains of the ligand protein. Owing to the fast internal rotation along the methyl axis, we theoretically and experimentally demonstrated the enhanced efficiency of this approach. The methyl-utilizing cross-saturation experiment has clear advantages in sensitivity and saturation transfer efficiency over the amide proton-based approach.

Rapid preparation of stable isotope labeled peptides that bind to target proteins by a phage library system.

We have developed a system for directly isolating foreign peptides displayed on the N-terminus of the major coat protein of bacteriophage M13. The phage particle in this system is formed as a mixture of wild type and modified coat proteins. The N-terminal segment of the modified coat protein was mutated for chemical cleavage, in order to obtain the displayed peptide from the major coat protein. Using 13C, 15N- labeled medium, we introduced stable isotopes, 13C and/or 15N, into the coat proteins. The NMR spectra for the cleaved peptides from the phage particles could be recorded within a few days after the selection of the phage clone.