Comprehensive Summary
This paper investigates Doublecortin-like kinase 3 (DCLK3). DCLK3 is a protein kinase that has been implicated in neuroprotection in Huntington’s Disease and other neurodegenerative diseases. DCLK3 has two well known paralogs: DCLK1 and DCLK2, however, DCLK3 itself is lesser understood and distinguished as a member of the “dark kinome”, a group of poorly characterized kinases. In order to begin to characterize DCLK3, the researchers used Phosformer-ST, an AI model designed to predict phosphorylation sites. The model identified a microtubule-associated protein, Tau, as a likely substrate for DCLK3. Using in vitro assays and mass spectrometry, Tau was confirmed to be an acceptable substrate, showcasing Phosformer-ST’s capability to identify relevant interactions. In order to develop potential structural models of DCLK3, AlphaFold 3 was used, an AI powered system that predicts 3-D structures of proteins. The use of AlphaFold 3 provided insight into the mechanism of how DCLK3’s autophosphorylation affects its structure.
Outcomes and Implications
This research enhances our understanding of the function and structure of DCLK3 through the use of AI models. Phosformer-ST and AlphaFold 3 have demonstrated their capability in exploring lesser-characterized proteins. This makes them invaluable tools in the field while offering a promising pathway for future research, enabling us to deepen our knowledge of protein biology.