Discovery of human IgG Fc-fragment binding ligands

The project concerns the discovery of small molecules binding selectively to the Fc fragment of human IgG, for use in affinity chromatography. The demand for alternatives to the currently used and very expensive Protein A is increasing with the number of monoclonal antibody (mAb) biopharmaceuticals entering clinical trials and reaching the market. The demand for alternatives are based partly on price considerations and partly on concerns about Protein A contamination of the biopharmaceutical. Here, we describe a rational combinatorial approach to discovery of novel small molecule ligands binding to the conserved Fc-Fragment of human IgG.

A dataset consisting of approximately 230.000 virtual ligands, comprised of a central scaffold to which two- or three building blocks were coupled, was generated. For each ligand, 118 Volsurf descriptors were calculated, and a Principal Component Analysis (PCA) was carried out to drastically reduce the number of latent variables. By applying D-Optimal Onion Design, a diverse subset of 291 ligands was identified. Another subset of 99 ligands was selected based on the proximity to 28 previously patented ligands. Chemical building blocks for the synthesis of a combinatorial library of ligands were selected based on the frequency of their occurrence in the selected ligands. The synthesized combinatorial library comprised 770 unique molecules.

The ligands were synthesized by employing the solid phase synthesis split and mix procedure.  By using optically encoded polymer beads for the solid phase synthesis, direct identification of the ligand composition was made possible. By reading the optical code of the beads in each reactor during the combinatorial synthesis, the composition of the ligand on each bead could be identified at any time. After the last step in the combinatorial ligand synthesis, the beads bearing the ligands were incubated with fluorescently labelled protein, and the fluorescence of each bead was quantified together with the bead code. This provided both the ligand structure and a measure of how well each ligand binds the target protein. As a result QSAR data was obtained for 88% of all ligands in the library; an unprecedented amount of information obtained from the screening of a single combinatorial library.

The building block composition of the identified high-affinity ligands was analyzed. Strong tendencies were observed, especially with respect to the building block coupled in the last step. It was attempted to construct a quantitative structure-activity relationship (QSAR) correlating the VolSurf descriptors with the measured fluorescence values, using partial least square regression to latent structures (PLS). Strong correlations were observed when only the molecules containing the most overrepresented building block 3 were used for the QSAR generation. However, it was not possible to construct a predictive QSAR from the descriptors and the measured fluorescence values, possibly due to lack of diversity of the molecules.

Internal Supervisors:
Professor Flemming Steen Jørgensen (FARMA)
Associate Professor Lars Olsen (FARMA)
Professor Thomas Bjørnholm (NSC)

External Supervisor:
Principal Scientist Phaedria Marie St. Hilaire (Novo Nordisk A/S)

Censor:
Professor John Nielsen (KU-LIFE)