IN SILICO IDENTIFICATION OF PUTATITVE CLAUDIN CHANNEL BLOCKERS

Abstract Compromised epithelial barrier function is known to be associated with inflammatory bowel disease (IBD) and may contribute to disease development. One mechanism of barrier dysfunction is increased expression of paracellular tight junction ion and water channels formed by claudins. Claudin-2 and -15 are two such channels. We hypothesize that blocking these channels could be a viable therapeutic approach to treat diarrhea in IBD. In an effort to develop blockers of these channels, we turn to our previously developed and validated in silico models of claudin-15 (Samanta et al. 2018). We reasoned that molecules that can bind with the interior of claudin pores can limit paracellular water and ion flux. Thus, we used docking algorithms to search for putative drugs that bind in the claudin-15 pore. AutoDock Vina (Scripps Research Institute) was initially used to assess rigid docking using small molecule ligand databases. The ligands were analyzed based on binding affinity to the pore and visualized using VMD (University of Illinois at Urbana-Champaign) for their potential blockage of the channel. Overall, a total of eight candidate ligands from the databases were identified: three from the UICentre database of 10000 ligands, one chemically similar structure identified in another online database (Chemspider), and four which are modifications on the chemical structure generated using ChemDraw. The analysis revealed that the eight ligands were docked in two predominant positions. In the first position, the ligands with more rings docked in an almost linear fashion and interacted with both D55 and D64 pore residues. In the second position of binding, the ligands were more flexible and could hence fold to interact only with D55 residues, thus biding predominantly in the center of the pores. To further evaluate these ligands, we will now turn to 1) flexible claudin-15 docking studies, 2) molecular dynamic simulations and, 3) in vitro measurements using monolayers induced to express claudin -15 and claudin-15 mutants. We also developed a claudin-2 homology model on which we will perform docking studies and in vitro measurements, which we expect will result in similar candidate ligands for blocking claudin-2. Finally, other databases will be analyzed for potential ligand blockers of claudin-2 and -15.