Abstract A021: In silico Prediction of Proteins Interacting with a Basal-like Pancreatic Cancer marker, RhoV, Using AlphaFold

Introduction Basal-like pancreatic cancer is a subtype with a worse clinical outcome than other subtypes. Ras homolog family member V (RhoV) is one of the basal-like markers that was recently identified. RhoV is a Rho GTPase and is closely regulated by guanine exchange factors (GEFs) and GTPase activating proteins (GAPs). An in silico prediction was performed using AlphaFold to identify RhoV regulators. Methods Amino acid sequences of 120 known GEFs and GAPs were collected. AlphaFold Colab, which uses AlphaFold2 and AlphaFold2-multier, was initially used to screen GEFs and GAPs for their binding affinities to RhoV. The top hits from AlphaFold2 were further verified by AlphaFold3, the most current version of AlphaFold partially open to nonprofit users. RAC1, a well-studied Rho GTPase with known GEFs and GAPs, was used as a control for the prediction process. Results AlphaFold predicted strong interactions of RAC1 and RhoV with GTP and GDP (ipTM of 0.92 between RAC1 and GTP; 0.95 between RAC1 and GDP; 0.92 between GTP and RhoV; 0.95 between GDP and RhoV). Known RAC1 binding partners, ARHGEF5 and ARHGAP21, showed strong binding affinities with RAC1 with ipTM of 0.91 and 0.92, respectively. Out of 120 GEFs and GAPs, AlphaFold2 predicted 6 of them with high binding affinities to RhoV (ipTM greater than 0.79). They are ARHGEF5, ARHGEF3, ARHGAP21, ITSN1, and ITSN2. Their binding affinities were confirmed by AlphaFold3, showing a strong correlation with AlphaFold2 (R of 0.79). AlphaFold3 also demonstrated high binding affinities between RhoV and compounds Palmitic acid (PLM), Citric acid (CIT), and Oleic acid (OLA). Conclusion AlphaFold is a revolutionary protein structural prediction tool with the ability to predict protein-protein binding affinities with high accuracy. AlphaFold2 is open to researchers in the field and allows unlimited queries on its server. AlphaFold3 only allows 20 predictions per research account per day. AlphaFold successfully predicted known binding partners of RAC1 and that RhoV could interact with the following GEFs and GAPs, ARHGEF5, ARHGAP21, and ARHGEF3, with high affinities. AlphaFold is a reliable in silico tool to screen binding partners of any protein of interest before meaningful biochemical interaction is confirmed in vitro and in vivo. Citation Format: Zebang Mark Li, Xiaoxu Zhou, Shaolei Lu. In silico Prediction of Proteins Interacting with a Basal-like Pancreatic Cancer marker, RhoV, Using AlphaFold [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research; 2024 Sep 15-18; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(17 Suppl_2):Abstract nr A021.