Detection of Drug-Protein Interactions with an Improved Yeast Three-Hybrid System

The identification of the targets of therapeutic drugs is one of the major challenges for pharmaceutical companies. The main target responsible for drug action is often known, but most drugs also bind other targets, either causing deleterious side effects in patients, or enabling the drug to be used for treating other diseases as well. Although plenty of methods are currently available for drug profiling, there are still a lot of drugs with unexplained side effects and with potential alternative uses. In this context, a novel method for target deconvolution was developed in the research group of Prof. Kai Johnsson, and is called Yeast Three-Hybrid. This method relies on the SNAP-tag technology for anchoring of the drug, and consists in screening libraries of human proteins for potential targets of drugs of interest. This novel approach was successfully applied in the lab and already led to the discovery of the mechanism of action of the anti-inflammatory drug sulfasalazine. In the first part of this work, automation of certain steps in the Y3H screening platform was implemented. This enabled more drugs and more libraries to be used at the same time. The automated platform was used to screen up to 12 libraries against 17 drugs of interest. Targets were then evaluated through methods based on affinity chromatography and protein cross-linking. Both methods rely on the SNAP-tag technology and therefore enable the same drug derivative to be used in both screenings and validation. The most relevant finding was that interactions between drugs and secreted proteins could be identified in screenings, although Yeast Three-Hybrid interactions occur in the nucleus of yeast cells. This is a significant result because it broadens the scope of the Yeast Three-Hybrid approach. In the second part of the project, the validation of Yeast Three-Hybrid hits and the sensitivity of the system were improved in a couple of ways. One important contribution was the development of a competition assay for easy validation of targets in yeast cells. This assay could definitely help make a selection if many hits were found in screenings and it would be necessary to decide with which hits to pursue further. In addition, the sensitivity of our system was compared to other Yeast Three- Hybrid systems. A comparison with the same system based on bacterial dihydrofolate reductase as bait yielded striking results for the drugs erlotinib and sulfasalazine. Taken together, the results showed that the dihydrofolate reductase-based system was more sensitive than the SNAP-based system, most likely because of the better permeability of the drug derivatives. Then, if only one system had to be selected for future screenings, it would most probably be the dihydrofolate reductase-based system. These last results were the most significant of this work because they initiated an important transition in the way future screenings will be performed. In summary, this work has brought many improvements to the Yeast Three-Hybrid screening platform in terms of throughput, hit validation, and sensitivity. These improvements will increase the chance of identifying protein targets of drugs in future screenings, and therefore contribute to the scientific progress in the field of drug profiling and more generally drug discovery.

Real-time binding kinetics of ligand-receptor interactions

Ekaterina Rostova

In principle, a major goal in drug development is to design a high-affinity compound with an appropriate pharmacological response, which requires understanding of physical processes underlying the drug-target interaction, affinity and specificity. This drug key information can be obtained by measuring drug binding kinetics in real time as the quantity of a ligand binding to a receptor immobilized on a biosensor chip surface. Almost all drugs independently of their targets interact with a complex cell membrane that incorporate various transmembrane proteins whose impact cannot be disregarded. Therefore, it is highly desirable that preclinical drug development is performed under conditions that are as close as possible to a native cell environment. The above reasons enhance the need for alternative approaches, such as whole-cell assays that are able to examine in vitro potential drug compounds. However, nowadays fragments of cell membranes have to be routinely isolated from the native cell environment due to limitations of currently available label-free technologies. In recent years, attempts were made to measure binding kinetics on living bacteria by surface plasmon resonance-based biosensors. Nevertheless, the experiments were not fully successful because of an insufficient penetration depth of the sensing evanescent field into living bacteria. Label-free biosensors based on metal-free photonic crystals support electromagnetic surface waves with a deeper penetration depth into the sample medium offering a solution for measuring binding interactions with living bacterial and, possibly, also eukaryotic cells. The deeper penetration of the surface waves creates a significant advantage for this non-disruptive technology over conventional immunoassays such as ELISA. For these reasons, label-free biosensors have been attracting a growing interest from pharmaceutical companies and research institutions by avoiding any experimental artefacts induced by labeling and allowing for real-time kinetic measurements. In this work, I contributed to the development of a label-free optical biosensor based on a photonic crystal that can be exploited for living bacteria. Moreover, I designed an experimental protocol and demonstrated how binding kinetics of various ligands, including antibodies and bacteriophages can be measured by the biosensor in vitro. Furthermore, this study attempts to apply the biosensor technology for bacterial viability tests exploring the ways in which the biosensor can detect metabolic activity of bacteria. Additionally, we performed theoretical studies on application of the Hillâs and Mathieuâs equations to study the effect of the photonic crystal design on the wave equation describing electromagnetic field distribution in the photonic crystal. This Thesis work demonstrates capabilities of a versatile biophysical method based on light interaction with living matter. In particular, the device sensitivity allows for measurements on living intact cells in vitro, which is a key technological advantage. Finally, this Thesis provides experimental evidence for future applications of the photonic crystal-based biosensor for antibody and bacteriophage therapies contributing to implementation of this technology in drug development process.

EPFL2017

Real-time binding kinetics of ligand-receptor interactions

Ekaterina Rostova

EPFL2017