Real-time drug discovery : characterization of the myelin-associated glycoprotein/ligand interaction

The challenge of drug discovery is the finding of potent molecules against relevant
targets that exhibit no or, in comparison to the benefit, tolerable adverse effects on
the health of patients. Costs for a new drug can be approximated to US$1 billion and
the overall process can last up to 12 years. For such a long-term investment, early
elimination of compounds with a poor outlook contributes significantly to cost and
time efficiency. Hence, it is fundamental to understand and predict pharmacological
as well as toxicological properties of potential drug candidates at early steps during
development.
After implementation of high-throughput screening in the 1990s the amount of
screened compounds per validated pharmacological target increased dramatically.
But higher throughput did not automatically result in larger numbers of new
molecular entities. A possible explanation is that current thinking and the
methodologies used in drug discovery mainly focus on equilibrium aspects of drug
binding (KD, IC50,…). In contrast, many successful drugs act in a non-equilibrium
way caused by very slow complex dissociation, covalent bond formation or based on
biochemical target properties e.g. a two-state model prolonging the binary complex
residence time. Whereas target properties are preset, the complex dissociation time
based on ligand properties is modifiable and drawing increasing interest among drug
researchers.
Despite the emerging use of techniques to determine interaction kinetics, no
consistent correlation between structural properties of ligands and kinetics was
published. This lack of knowledge complicates modification of lead structures
towards favourable kinetic properties. Currently, measuring kinetics only provides
the possibility to differentiate between e.g. slow and fast dissociating leads.
The topic of the thesis was to implement surface plasmon resonance (SPR) based
biosensors to determine equilibrium binding constants, kinetic rate constants and
enthalpic/entropic contributions of ligands binding to the myelin-associated
glycoprotein (MAG). In an interdisciplinary environment with chemists, biologists,
NMR and molecular modelling specialists, interaction data was required to support
the finding of new leads and ligands exhibiting advanced binding properties.
The SPR-based biosensor was validated on a model system for carbohydrate/protein
interactions. A diagnostic anti-sialyl Lewisa (sLea) antibody (GSLA-2) used in tumor
diagnostic was investigated. High selectivity of GSLA-2 towards the sLea epitope
was found and a close interaction with STD-NMR finally enabled an exact
description of the binding epitope. Additionally, the influence on interactions caused
by hydrophobic labels attached to a ligand was described.
For the myelin-associated glycoprotein, comparison of several immobilization
approaches displayed specific differences important for the characterization of
drug/target interactions. On all surfaces, reproducible results were obtained in good
agreement with data from a hapten inhibition assay. Interesting though unusual was
the occurrence of negative sensorgrams on surfaces with oriented immobilization
(capturing and enzymatic). These were only observed for sialic acid derivatives, a
certain class of MAG ligands. Larger molecules like the ganglioside GT1b or secondsite
ligands displayed positive signals. Supporting experiments proved the negative
signals to be specifically derived from ligand binding. The most reasonable
explanation was a ligand-induced conformational change caused by sialic acid
derivatives.
Whereas capturing led to a baseline drift, enzymatic immobilization improved the
assay format significantly. Implementation of enzymatic immobilization provided a
highly reliable technique. Practicability for very low protein concentrations (4 μg/ml)
facilitated a direct immobilization of secreted proteins out of cell culture supernatant
delivering a gentle procedure.
All sialic acid-based ligands displayed fast association and dissociation rate constants
with a complex half-life in the range of 1-2 seconds. Therefore, future search for new
leads exhibiting slower dissociation appears promising. Interactions of sialic acid
derivatives with MAG were found to be mainly enthalpy driven. A decent correlation
between structural modifications and their enthalpic/entropic contribution was found.
Furthermore, it was possible to support and even expand data published on signal
deviations related to drug-like molecules.
The non-availability of the crystal structure of the myelin-associated glycoprotein
complicated rational ligand design but was successfully compensated by a
combination of SPR-based biosensor data, STD-NMR and homology modeling. Even
though a crystal structure would significantly simplify ligand design, a detailed
picture of the interaction process including important kinetic data could be obtained.