This work was performed by a collaborator’s lab. My contribution to it was only one of the core ideas behind the work and some of the bioinformatics-based exploration, leading to the initiation of the work. So, I shall explain that here. Erythropoietin (EPO) is active in both the glycosylated (natural) and non-glycosylated (recombinant; bacterially-produced) forms; however, when the non-glycosylated form is injected into the organism it is very quickly cleared. I suggested that this could be because the sugar residues act to protect certain nearby peptide bonds on the surface of the protein from undergoing proteolytic cleavage, by acting as steric blockers of the approach of proteases. Perhaps, I argued, the lack of the sugars (or their removal) is designed to be associated with rapid proteolytic cleavage, even as a natural clearance mechanism. Would it be possible to test this idea through structural bioinformatics approaches? If the idea appeared to have some validity, could one then design EPO variants that would resist proteolysis? To answer these questions, we looked at the solvent exposure of the C=O and N-H atoms of every peptide bond in EPO without any prejudice to identify the ones that have the maximum solvent exposure and, therefore, the maximum likelihood of undergoing proteolytic cleavage. Separately, we collected available information about the sites on EPO’s surface which are known to be glycosylated. We discovered a very remarkable correlation between the sites of glycosylation and the most solvent-exposed peptide bonds; it appeared almost as if the sugar chains could potentially be ‘rotating about and acting like fans’ in the vicinity of the peptide bonds, sterically preventing the approach of proteases. To now protect the exposed bonds, we examined the neighbouring side-chains flanking such potentially ‘labile’ peptide bonds and found that they consisted of very few atoms (small side-chains). Since none of these sites were important for receptor binding, we designed various schemes of replacement mutations, to fortify these peptide bonds by flanking them with larger side chains conserving the nature (but not the identity) of flanking residues. Many of these designer mutations were made and tested by the collaborating group, and all the data in the paper is derived from the work of this group.

Click to Read in Detail