Structural and molecular determinants affecting the interaction of retinol with human CRBPs

Four intracellular retinol-binding protein (CRBP) isoforms, belonging to the intracellular lipid-binding protein (iLBPs) family, have been reported in human. Despite their low sequence identity, they show a high structural conservation. CRBP 1 and 2, for which retinol is the endogenous ligand, bind retinol with high affinity, while CRBP3 and 4 show for it a low and very low affinity, respectively. The absorption spectra of the retinol bound to CRBP1, 2 and 3 shows a peak with a maximum at approximately 348 nm and two shoulders, while the same assay on CRBP4 shows a different spectrum with a maximum at about 330 nm. One of the most obvious differences between the four isoforms is a substitution in one of the two binding key residues: the glutamine 108, present in CRBP1 and 2, is replaced with a histidine residue in CRBP3 and 4. In this work we have studied the role of the key binding residues by the characterization of three CRBP1 mutants: K40L, Q108L and the K40L/Q108L double mutant, demonstrating that the substitution of both the key residues preserves the ligand up-take and that hydrophobic interactions between hydrophobic residues of the cavity and the lipophilic part of retinol are sufficient to hold the ligand inside the pocket. Therefore, other factors, distinct from the substitution of one of the binding residues, are responsible for the differences in the behavior of CRBPs. In addition, our data suggest the presence of a cation ?-interaction between the side chain of Lys40 and the retinol poly-isoprene tail. The different flexibility of the four isoforms could be an important feature to clarify CRBP differences. To this aim, we have performed MD simulations on the apo-proteins, observing a higher rigidity of CRBP1 and 2 compared to the flexibility of CRBP3 and 4. Different regions belonging to distinct isoforms are interested by different mobility. The ligand up-take mechanism of iLBPs is a broadly debated topic and for CRBPs is not yet understood. We have carried out MD simulations on the four isoforms in the presence of retinol and we have proposed two different up-take models for the binding mechanism of CRBP1 and 2. The ligand up-take simulations on CRBP3 and 4 have suggested the possibility that these two isoforms could bind compounds different from retinol. Finally, we have studied the interaction of CRBP1, 3 and 4 with a new natural retinoid fortuitously identified during a protein purification passage.