February 21, 2018Download PDF
BioSight* is the BIONEXT’s Web interface allowing the retrieval and assessment of local surface similarities. Here, BioSight was challenged with two serine/threonine kinase binding pockets. Top ranked proteins are mainly kinases, of course. But some other relevant proteins, not belonging to the kinase family, are easily found by BioSight. These data could help unravel parts the pharmacological profile for some kinase inhibitors.
(*) Create your profile on BioSight, the fully secured Web access to BioBind, to access the full analysis
Kinases arouse great interest among the pharmaceutical society. As key enzymes for cell life, they are targeted for the treatment of various diseases such as cancer, diabetes or inflammation. Here, we take a particular interest in serine/ threonine kinases (EC 22.214.171.124), enzymes that phosphorylate serine and threonine. The proto-oncogene serine/threonine protein kinase (PIM-1) regulates many cellular processes including cell cycle progression and apoptosis. It is also involved in prostate cancer, pancreatic cancer or sporadic malignant tumors. The mitogen-activated protein kinase 12 (MAPK 12) is a putative therapeutic target in the treatment for rheumatoid arthritis, psoriasis and inflammatory bowel disease. Starting from these two, the present test case challenges the ability of the BioSight Web platform to retrieve secondary targets not belonging to the kinase family, and assesses their relevance.
BioSight embeddes BioBind1, a BIONEXT’s patented application developed for the detection and ranking of similar sites in the study of biological macromolecules2. The query surface pattern is defined using atoms of the macromolecule nearby a given ligand and is compared to all crystallized structures contained in the PDB database (> 35,000 proteins). Two kinase structures have been used for this challenge: PIM-1 kinase (PIM1_HUMAN, PDB 3CY3) in complex with the JNK inhibitor V in the ATP binding site and the Mitogen-activated protein kinase p38 gamma (MK12_HUMAN, PDB 1CM8) in complex with ANP. The query sites have been defined using protein atoms around the adenin key substructure, in order to get the most specific query surface as possible.
As shown in Figure 2, a plot of the score distribution for the PIM-1 query, the first ranked structures are all kinase proteins, attesting that the characteristic fold of kinase is well retrieved by BioBind.
Here we will focus on ATP binding proteins other than kinases. BioBind spotted some really interesting similarities with the que- ry. We report three examples below (see Table 1). For each of these proteins, BioBind correctly identified the ATP binding pocket and found a relevant similarity with the query. As shown in Figure 3 with the molecular chaperon HSP70-1, the ligand from the query is well superposed to the ADP from the target, highlighting that the similar surfaces found by BioBind correspond to equivalent interaction mode.
For the MAPK 12 query, BioBind preferentially retrieved kinases, too. BioBind also found relevant similar surfaces amongst the non kinase proteins (see Table 2, Figures 1,4 & 5).
Even more relevant, synapsin-3 is the first non kinase protein retrieved by BioBind. Its global fold is different from the query but BioBind highlighted a really similar binding site, supported by very locally similar secondary structures (see Fig.5). Furthermore, the inhibition of synapsin using kinase inhibitors have indeed been proved experimentally3.
Using the BioSight platform, we successfully challenged our algorithm BioBind on two different serine/threonine kinases queries. As expected, for both queries, BioBind retrieved predominantly kinases among the first hits. Nonetheless, among the other highly ranked results, several relevant proteins were found (ranked 1, 9 and 30, excluding kinases, for the MAPK 12 query), for which the similarity is justified by comparable interaction modes of the ligands. This showcase reveals the potency of BioSight to detect sites with similarity to a given query even outside the query family. This ability is particularly useful in a pharmaceutical context, for instance for the prediction of secondary targets or drug repurposing. Our fully secured platform can be accessed and challenged at https://biosight.bionext.com
- Bionext’s private publication, available on demand. [return]
- Albou l.p., et al. (2009). defining and characterizing protein surface using alpha shapes. proteins: structure, function, and Bioinformatics, 76(1), 1-12. [return]
- De franchi, e. et al. (2010). Binding of protein kinase inhiBitors to synapsin inferred from pairwise Binding site similarity measurements. plos one, 5(8), e12214. [return]