Abstract:
Fragment-based drug discovery is an efficient approach to probe the chemical space of protein binding sites. Screening experiments can yield fragment hits that can be further investigated.
By designing focused libraries one can try to shift binding modes towards underrepresented
or unconventional binding interactions. One such underrepresented interaction is the σ-hole
interaction of e.g., halogens or chalcogens. Halogen-containing molecules that play biologically or pharmaceutically relevant roles often have the halogen bound to either an aryl or alkyl group.
Contrary to common perception, halogens (Cl, Br, and I) do not exhibit an isotropically distributed electron density, but rather an anisotropically distributed one. Along the C–X-axis a positive electrostatic potential is observed, while a negative electrostatic potential wraps around the bond. This phenomenon allows for halogens to closely interact with electronegative groups such as carbonyl oxygen atoms and many more. Previously, the group of Prof. Frank Böckler has designed such a halogen-focused library, called the halogen enriched fragment library (HEFLib).
One important lesson that was learned from characterizing this fragment library was that the size of the σ-hole, given as the Vmax-value, correlates with hit rates.
In addition, the knowledge about the exact atomic coordinates is fundamental to knowing
the exact binding location on the protein, if a σ-hole interaction is present, if and how the
hit can be optimized, or if a structural change in the protein occurred. X-ray crystallography
is used to elucidate the binding mode of a ligand–protein complex by obtaining atomic coordinates.
Herein, I present the design of a Vmax-value optimized halogen enriched fragment library
(Vmax-Lib). Based on hits obtained from this and complementary designed libraries (HEFLib,
Cov-Lib, and the Chalco-Lib) on multiple targets, I solved a total of 31 crystal structures with
these ligands bound to the respective targets. These include ten structures of mutant p53,
two of DYRK1a, 17 of JNK3 or JNK3-M115L, one of USP7 and one of cereblon. Further,
I investigated reasons for selectivity between JNK2 and JNK3 and focused on the question
if a methionine–methionine contact could be a major contributor. Additionally, some crystal
structures have shown unexpected features that will be presented, such as a possible unreported NOS bond in JNK3 or a possible unique rescue mechanism for mutant p53-R282W. Finally, I explored if an expression and purification protocol for S6K2 in E. coli can be established.
While expression of S6K2 was possible issues with solubility and aggregation did not allow for a successful purification, suggesting that other expression systems may be necessary.
Future work in obtaining more ligand–protein complexes is important for rational drug design,
and can identify binding modes suitable for fragment growing, merging or linking. More
structures of ligands binding to both JNK3 and JNK3-M115L is necessary to better understand the effect of the Met–Met contact. All in all, the obtained structures enabled us to rationally pursue structure-activity relationship trials, confirmed halogen bonding and revealed unexpected features that paved the way for further investigation.
Halogen Bonding