Combining SAXS with Rosetta for Identification of Native-like Protein Complexes

It is well-known that recombinant human serum albumin (rHSA) has the ability to stabilize proteins in solution preventing protein adsorption, aggregation and oxidation. For this reason, rHSA is used as an excipient in the formulation of protein pharmaceuticals. To shed light on the molecular interactions, we have studied a variety of protein drugs that are known to bind to rHSA and thereby being stabilized. We observe that the interactions depend on protein concentrations and differ significantly between protein drugs. One approach to study these systems on a molecular level is the combination of small angle X-ray scattering (SAXS) and in-silico modelling. SAXS can be used to identify the overall shape of proteins and protein complexes and ab initio models can be derived from the scattering profile using programs such as Dammif [1]. These programs allow us to assess the overall conformation of the macromolecular structures, but cannot provide detailed information on the molecular level regarding protein-protein interfaces of the complexes. Here, the Rosetta modelling suite, a multipurpose software suite, can be utilized to perform protein-protein docking and to study the complexes. The challenge in using the Rosetta docking tool [2] is the difficulties in efficiently identifying the native-like structure. For better identification we apply SAXS constrains during the docking procedure. Although the method has been applied previously [3], no benchmarking has been published regarding the relative success of using SAXS constrains. We therefore have conducted an elaborate benchmarking, where we have used SAXS constrains for determination of complexes of non-identical components using the Rosetta docking protocol. A pool of complex structures has been chosen to evaluate the difference between conventional docking and docking performed using SAXS constrains. This allows us to optimize the parameters in the protocol and pave the way to study unknown complex structures.

The impact of zinc on insulin fibrillation

Protein and peptide amyloid fibrillation is an increasingly significant field of research due to the growing prevalence and recognition of amyloid-related human diseases and in relation to therapeutic applications of peptide drugs where it constitutes a challenge during production and transportation. Insulin fibrillation represents a good model system for fibrillation studies and numerous pathways of insulin fibrillation have been suggested [1]. In general, partially unfolded insulin monomers (or dimers) are believed to be a vital prerequisite for prefibrillar association and eventually fibril formation (reviewed by Groenning in [1]). As fibrillation is commonly found to occur from the biologically active, monomeric form of insulin, zinc is considered to impede insulin fibril formation through stabilization of the physiological predominant hexameric forms. The question arises whether the assigned monomeric pathway at least to some degree could be a premise generated from the usual experimental procedure of inducing fibril formation; experiments that are conducted at acidic conditions with protonated histidine residues that cannot coordinate the Zn(II) ions, to induce the monomeric form believed to be the prerequisite of insulin fibrillation? However, in a more recent study [2], it was suggested that Zn(II) ions in contrast also inhibit fibrillation through differential stabilization of the insulin monomer. In the present work, we have investigated the presence and influence of zinc at well-defined stoichiometric levels at physiological pH, as well as the inner coordination sphere of Zn(II) during insulin fibril formation, using X-ray absorption spectroscopy (XAS). In addition, the results were validated with fiber diffraction studies, small-angle X-ray studies and ThT fluorescence studies.

X-ray powder diffraction: A powerful tool for industrial protein production

X-ray powder diffraction (XRPD) offers a method of characterizing a crystalline protein suspension [1], and data can be collected within 30 minutes, which is appealing for industrial applications. In industry, enzymes are produced and handled in high concentrations, which can in turn cause problems for the processes due to protein precipitation in the production pipeline. XRPD is useful for identification of the crystal forms present, by fitting calculated patterns of known single crystal forms to the observed XRPD pattern. For this purpose we have developed a streamlined program for calculation of diffraction patterns from pdb-files taking into account bulk-solvent, peak asymmetry and background [2]. XRPD was applied to a suspension from a large-scale industrial production of the widely used Bacillus lentus subtilisin. A dominant crystal form was identified by XRPD, but two other different crystal forms were found by a complementary single crystal micro-diffraction analysis of the larger single crystals present in the sample [3]. The study serves as a reminder that when a crystal is picked out from a batch crystallization for single crystal analysis, it might not be representative of the bulk microcrystalline material in the sample. To estimate the fraction of the different crystal forms in production samples with significant polymorphism, a further XRPD study was performed on binary mixtures of different lysozyme and subtilisin crystal forms. Quantitative XRPD generally requires careful sample preparation, and working with protein slurries leads to further challenges in terms of varying crystal density. After careful optimisation of suspension medium, the relative composition of crystal forms can be determined within 10%. This work demonstrates the value of in-house XRPD as an analysis tool in industrial enzyme production, and its potential to help troubleshooting the production process and to provide information for further refining the manufacturing of enzymes.