Identification of molecular determinants of CXCR3-gliadin-mediated triggering of intestinal permeability

CXCR3 is a G-protein coupled receptor expressed principally on leukocytes, monocytes and epithelial cells; it is involved in leukocyte traffic, integrin activation, cytoskeletal changes and chemotactic migration by binding to its classical ligands, CXCL-9/10/11. Moreover, it is involved in celiac disease by binding of 2 peptides produced from gliadin digestion (111-130 and 151-170). This interaction induces an increase of intestinal permeability in a zonulin dependent way: the cytosolic adapter protein MyD88, crucial for the maintenance of gut homeostasis, is recruited and activates zonulin release. The latter, in turn, transactivates epidermal growth factor receptor (EGFr) through proteinase activated receptor 2 (PAR2) leading to tight junctions disassembly by the combination of TJ protein phosphorylation and actin polymerization. This causes the rearrangement of the filaments of actin and the subsequent displacement of proteins from the junctional complex. Three splicing variants of CXCR3 are known: CXCR3-A, the most common isoform, consists of 368 amino acid residues and is the most frequently expressed on immune cells; it couples to a G-protein which mediates pro-migratory and proliferative signalling and increases intracellular calcium levels; CXCR3-B, which can bind CXCL4 in addition to classical CXCL, results from an alternative splicing of the CXCR3 mRNA with a 52 aa extended NH2-terminal domain when compared to the isoform A, while CXCR3-alt is a significantly truncated variant activated only by CXCL11 and not involved in celiac disease. Recently it has been demonstrated that isoform A is more abundantly expressed in the intestinal mucosa of celiac patients, while the B isoform in gluten sensitive, non-celiac patients suggesting that both isoforms are involved in gliadin binding. Our goal is to evaluate the differential binding of the natural ligand, CXCL10 (or IP-10), and of the two gliadin peptides, on the two CXCR3 isoforms (-A and -B) involved in intestinal permeability. The 3D model of both isoforms, CXCR3-A (UNIPROT-id: P49682-1) and CXCR3-B (UNIPROT-id: P49682-2) were obtained from GPCR-I-TASSER [1]. Models were included in a membrane system with CHARMM-GUI server [2]. The obtained complexes were refined by MD simulation in agreement with CHARMM-GUI suggested protocol for Gromacs. X-ray structure of IP-10 was obtained from PDB (PDB ID: 1LV9), while gliadin peptides were predicted using different de novo peptide structure modeling servers: PEP-FOLD 3 [3] , PEPstrMOD [4] and QUARK [5]. Overall, 4 conformers were obtained for peptide 111-130 and 5 for 151-170. Protein-protein docking of CXCR3(A/B)-IP10 was performed with Haddock [6] and ZDOCK [7] servers, as well as protein-peptide docking of CXCR3(A/B)-gliadin(111-130/151-170) was performed with Z-DOCK, Haddock, and a coarse grained (CG) application of AutoDock. CXCR3(A/B)-IP10 docking outputs from ZDOCK have been discarded as the ligand did not interact with the binding cavity of the receptor, while Haddock results have been evaluated through structural analysis (FCC, i-RMSD and l-RMSD) and the HADDOCK model with the lowest score has been selected for further simulations. Docking complexes of CXCR3(A/B)-gliadin peptides obtained from ZDOCK have been clustered by chimera clustering tool and 1 conformer for each complex has been achieved; Moreover, HADDOCK run returned 1 or 2 clusters for each peptide-protein docking simulation. The AutoDock CG simulation on CXCR3(A/B)-gliadin peptides is ongoing. Outputs will be processed in order to compare the different docking algorithms employed. Given that our aim is to identify the molecular determinants of peptide binding instead of the natural ligand and the different binding mode and affinity between CXCR3 isoforms, complexes will undergo MD simulations. Moreover an MMPBSA analysis on the trajectories will be performed in order to determine the binding affinity.