We selected candidate for experimental test using the principle that mutant should bind the receptor roughly as well as wild type for the better-binding 〖His〗^° tautomer (corresponding to cell-surface conditions) and substantially more poorly than wild type for 〖His〗^+ (corresponding to endosomal condition). Once the difference between binding 〖His〗^° and 〖His〗^+ is more than a few kcal/mol , there is no adventage to making it larger because it would be greater than the cost of deprotonating 〖His〗^+ in the
unbound state and binding as 〖His〗^° . Using these guidelines, D110H and D113H mutant were made and purified for experimental characterization.
Experimental receptor-binding properties of wild type and mutants. Because there are no native
histindine residues in wild-type GCSF at the binding interface of interest, the wild type serves as an appropriate control in that it should show minimal sensitivity to binding affinity at endosomal pH relative to extracellular pH. To test whether our biochemical property targets were met, we carried out
equilibrium binding experiment for wild type and each of the Asp → His mtant at pH7.4 and pH5.5 (Table 1 and Fig. 1D-F). Both surface plasmon resonance and cell experiments demonstrate that there is no substantial difference in binding affinity between the mutant and wild-type ligand at pH7.4,
consistent with our molecular modeling perdictions. At the same time, the cell experiment demonstrste
that the histidine mutants exhibit a substantial decrease in binding affinity at pH5.5 relative to pH7.4
whereas wild-type binding remain largely insensitive to pH; this is also consistent with our computational modeling predictions. Because the mutants bind to GCSFR with substantially lower affinity than wild type under conditions representative of the endosomal pompartments, we could then predict a cell-level result of increased ligand lifetime in the medium for the histidine analogs relative to native GCSF due to enhanced recycling.