Identification of novel linear motif binding

partners through structure-based design

1 1
Gábor Erdős , Zsuzsanna Dosztányi

1
MTA-ELTE Momentum Bioinformatics Research Group

Department of Biochemistry, Eötvös Loránd University, Budapest

Keywords: Disordered proteins; Short linear motifs; Bioinformatic prediction
methods

Abstract
Short Linear Motifs (SLiMs) are compact functional modules that are recognized by specific
globular domains. Such interactions play important roles in many biological systems; they
regulate the formation of transient protein complexes, direct subcellular localization, and
determine the fate of proteins. Although general structure based methods exist for partner
prediction, they do not perform well for this specific linear motif mediated systems. Results
from AlphaFold2 show that while in some cases it is able to recognise linear motif mediated
interactions, it is not suitable for partner prediction. Sequence based methods also suffer
from various problems, as they cannot capture the global attributes of the binding.

In the presented work we designed a method which is able to identify new binding partners
for a given linear motif binding domain based on the structure and binding mechanism of its
complex. Our approach utilizes a unique statistical potential based force-field that was
specifically trained to identify known binding partners against randomly generated motifs.
The presented method is able to assess the global likelihood of a region of a protein to
contain a short linear motif, as well as give an indication to what domain that motif binds to.

We are testing this method in different systems such as LC8, SIAH, RB and P107.