Controlled dehydration, structural flexibility, and Gadolinium MRI contrast compound binding in human plasma glycoprotein afamin

Afamin, a human blood plasma glycoprotein, a putative multi-functional transporter of hydrophobic molecules and a marker for metabolic syndrome, poses multiple challenges for crystallographic structure determination, both practically and in analysis of the models. Several hundred crystals were analysed, and unusual variability in cell volume and difficulty solving the structure despite a ~34% sequence identity with non-glycosylated human serum albumin indicated that the molecule exhibits variable and context-sensitive packing, despite greatly simplified glycosylation in insect cell expressed recombinant afamin. Controlled dehydration of the crystals was able to stabilise the orthorhombic crystal form reducing the number of molecules in the asymmetric unit from the monoclinic form and changing the conformational states of the protein. An iterative strategy, using fully automatic experiments available on MASSIF-1, was used to quickly determine the optimal protocol to achieve the phase transition that should be readily applicable to many types of sample. The study also highlights the drawback of using a single crystallographic structure model for computation modelling purposes given that conformational state of the binding sites and electron density in the binding site, likely resulting from PEGs, greatly varies between models. This also holds for the analysis of unspecific low-affinity ligands, where often a variety fragments with similar uncertainty can be modelled, inviting interpretative bias. As a promiscuous transporter, afamin also seems to bind Gadoteridol, a magnetic resonance imaging contrast compound, in at least two sites. One pair of Gadoteridol molecules is located near the human albumin Sudlow-site, and a second Gadoteridol at an intermolecular site in proximity of domain IA. The data from the co-crystals provide an opportunity to evaluate modern metrics of data quality in the context of the information that can be gleaned from data sets that would be abandoned on classical measures.