Search

Raw data for the X-ray structure determination of a xenon-derivatized isoprenaline∙β1AR crystal, the xenon atom anomalous maps, NMR spectra, and for the ITC curves presented in the manuscript "Filling of a water-free void explains the allosteric regulation of the β1-adrenergic receptor by cholesterol". The main folder (Xe_isoprenaline_b1AR) contains two subfolders. The coordinate file (.pdb), the hkl file, and the electron density map (.mtz) are placed in the folder "structure". The second folder, “anomalous_map”, contains the anomalous maps (.pha and .map), and the hkl files for the four different crystals evaluated (organized in separate folders for the each crystal 1-4). The images for the 4 crystals are placed in separate folders (images_crystal_1, images_crystal_2, images_crystal_3, and images_crystal_4). The folder “NMR” contains processed 1H-15N TROSY NMR spectra (.DAT files), and full Bruker directories for 1H experiments. The datasets are organized in subfolders named according to the manuscript figures. The folder “ITC” contains the original ITC files (.itc) presented in Figure 4 and supplementary Figure S5 of the manuscript., This work was supported by the Swiss National Science Foundation (Grants CRSK-3_195592 to L.A.A., and 31-149927, 31-173089, 31-201270 to S.G.).

In eukaryotic cells, microtubules are highly dynamic filaments of the cytoskeleton where they are involved in vital tasks like cell division, intracellular transport, and cell polarization. Microtubules are assembled from their building block tubulin, and both microtubules and tubulin are known to bind a multitude of regulators. In cells, protein partners bind to microtubules to either modulate their behavior or to use them to fulfill various cellular functions. In addition to proteins, several dozen small molecule ligands are known to influence microtubule dynamics and they target six distinct binding sites on tubulin. As tumor cells exhibit faster rates of cell division than healthy cells, it is not surprising that mitotic inhibition is one of the many strategies to treat cancer. Indeed, five of these ligands are currently used in the clinic as monotherapeutic antitubulin drugs. While their mechanism of action differ, all original ligands are derived from natural sources and have complex chemical structures. Based on these observations, we aimed in the present thesis to address the two outstanding open questions whether there are additional binding sites present in tubulin and if it is possible to develop a small molecule ligand targeting microtubules. To tackle these two questions, we first performed an X-ray crystallography based fragment screen. Using this approach, we found 56 chemically diverse fragments that bound to 10 distinct binding sites on tubulin. Moreover, we combined a computational binding-site search with the results from our fragment screen to comprehensively map all binding sites in tubulin. Indeed, we found 18 novel sites that are not targeted by any of the known ligands. Out of these, 11 sites are also not targeted by any known tubulin protein partner. These findings demonstrate that our combined computational and crystallographic approach is a powerful tool to map binding sites in any protein for which a well diffracting crystal system is available. With the structural information at hands on a plethora of new fragments bound to tubulin, we went on to develop novel small molecule ligands targeting tubulin. As a first step and proof of principle, we chose a fragment that binds to the well-known colchicine site and has structural resemblance to nocodazole, a microtubule-destabilizing agent that was proven to be too toxic to use in the clinic. This fragment establishes an intricate interaction network with tubulin, consisting of four hydrogen bonds. Indeed, when probing for the importance of these hydrogen bonds by removing them individually we learned that all of them are necessary to retain binding. While leaving the interaction network unchanged, we still had the option to grow our fragment by attaching continuously larger substitutions and in this way, we managed to improve our fragment from an IC50 of 54 μM to a final IC50 of 0.94 μM in our best compound. These results demonstrate that already a small fragment can exhibit optimal interactions to its target and that indeed our fragment hits can be used to obtain potent tubulin inhibitors. As our pilot project showed that our fragments can be improved, we went on to apply our freshly gained insight to a novel tubulin-binding site. This time we chose a fragment linking strategy, in which several fragments that bind close to each other are combined. First, we took a fragment that contained a common binding motif as a starting point to gain specificity. Second, we looked for additional fragments binding in close proximity that could be chemically linked to our starting fragment. The two additional fragments that we selected suggested that expanding our starting fragment with a phenyl moiety should increase its affinity. Indeed, with our linking effort we obtained a first ligand with an IC50 of 48 μM, whereas none of the starting fragments showed any effect on cells. Additional rounds of modification of this first ligand brought us our most potent compound, which we named Todalam. Todalam has an IC50 value of 8.8 μM and showed microtubule depolymerizing effects in vitro as well as in cells. Furthermore, we proposed a novel, twofold mechanism of action for Todalam: First, Todalam acts like a “plug” locking tubulin in its curved conformation and second, it links free tubulin to ring like structures that are incompatible with microtubule assembly. This is to the best of our knowledge the first rationally designed tubulin-targeting agent with a novel binding site and a novel molecular mechanism of action. Overall, we demonstrated that crystallographic fragment screening is a very powerful tool both for finding novel binding sites as well as for developing small molecule ligands. Our results set a basis for future ligand design efforts by providing a diverse set of novel tubulin-binding sites and ligands. They further offer two lead-like compounds that are ready to be developed into small lead molecules.

35 p.-12 fig.-2 tab.-3 animat.-1 schem.This article is dedicated to the memory of Dr Linda Amos, a dear friend and pioneer in the study of microtubules and the mechanism of action of paclitaxel (Amos and Löwe, 1999), who passed away while we were assembling this manuscript., Paclitaxel (Taxol) is a taxane and a chemotherapeutic drug that stabilizes microtubules. While the interaction of paclitaxel with microtubules is well described, the lack of high-resolution structural information on a tubulin-taxane complex precludes a comprehensive description of the binding determinants that affect its mechanism of action. Here, we solved the crystal structure of baccatin III the core moiety of paclitaxel-tubulin complex at 1.9 Å resolution. Based on this information, we engineered taxanes with modified C13 side chains, solved their crystal structures in complex with tubulin, and analyzed their effects on microtubules (X-ray fiber diffraction), along with those of paclitaxel, docetaxel, and baccatin III. Further comparison of high-resolution structures and microtubules’ diffractions with the apo forms and molecular dynamics approaches allowed us to understand the consequences of taxane binding to tubulin in solution and under assembled conditions. The results sheds light on three main mechanistic questions: (1) taxanes bind better to microtubules than to tubulin because tubulin assembly is linked to a βM-loopconformational reorganization (otherwise occludes the access to the taxane site) and, bulky C13 side chains preferentially recognize the assembled conformational state; (2) the occupancy of the taxane site has no influence on the straightness of tubulin protofilaments and; (3) longitudinal expansion of the microtubule lattices arises from the accommodation of the taxane core within the site, a process that is no related to the microtubule stabilization (baccatin III is biochemically inactive). In conclusion, our combined experimental and computational approach allowed us to describe the tubulin-taxane interaction in atomic detail and assess the structural determinants for binding., Ministerio de Ciencia e Innovación (PID2019-104545RB-I00 (AEI/10.13039/501100011033)) J Fernando Díaz Consejo Superior de Investigaciones Científicas (PIE 201920E111) J Fernando Díaz Fundación Tatiana Pérez de Guzmán el Bueno (Proyecto de Investigación en Neurociencia 2020) J Fernando Díaz European Union NextGenerationEU (H2020-MSCA-ITN-2019 860070 TUBINTRAIN) Andrea E Prota J Fernando Díaz Swiss National Science Foundation (310030_192566) Michel O Steinmetz JSPS KAKENHI (16K07328/17H03668) Shinji Kamimura National Natural Science Foundation of China (30930108) Wei-Shuo Fang Chinese Academy of Medical Sciences (2016-I2M-1-010) Wei-Shuo Fang Ministerio de Ciencia e Innovación (PID2019-104070RB-C22 (AEI/10.13039/501100011033)) Federico Gago We thank Ganadería Fernando Díaz for calf brains supply and staff of beamlines X06DA of the Swiss Light Source (Paul Scherrer Institut, Villigen PSI, Switzerland) and BL11-NDC-SWEET (ALBA, Cerdanyola del Vallès, Spain) for their support. We also thank Mr Pedro Gascón Blanco for his private donation to the project to support a month of a student salary.