Search

A refolding reactor was developed for continuous matrix-assisted refolding of proteins. The reactor was composed of an annular chromatography system and an ultrafiltration system to recycle aggregated proteins produced during the refolding reaction. The feed solution containing the denatured protein was continuously fed to the rotating bed perfused with buffer promoting folding of the protein. As the protein passed through the column, it was separated from chaotropic and reducing agents and the refolding process took place. Native proteins and aggregates could be continuously separated due to different molecular size. The exit stream containing aggregates was collected, concentrated by ultrafiltration and recycled to the feed solution. The high concentrations of chaotropic and reducing agents in the feed solution enabled dissociation of the recycled aggregates and consequently were fed again to the refolding reactor. When the initial feed mixture of denatured protein is used up, only buffer-containing chaotropic agents and recycled aggregates are fully converted to native protein. This process resulted in a stoichiometric conversion from the denatured protein to its correctly folded native state. The system was tested with bovine alpha-lactalbumin as model protein. Superdex 75 PrepGrade was used as size-exclusion medium. The yield of 30% active monomer in the batch process was improved to 41% at a recycling rate of 65%. Assuming that the aggregates can be redissolved and recycled into the feed stream in a quantitative manner, a refolding yield close to 100% is possible. The method can be also applied to other chromatographic principles suited for the separation of aggregates.

Hydrophobic interaction chromatography media suited for large scale separations were compared regarding dynamic binding capacity, recovery and mass transfer properties. In all cases, pore diffusion was the rate limiting step. Reduced heights equivalent to a theoretical plate for bovine serum albumin derived from breakthrough curves at reduced velocities between 60 and 1500 ranged from 10 to 700. Pore diffusion coefficients were derived from pulse response experiments for the model proteins alpha-lactalbumin, lysozyme, beta-lactoglobulin, bovine serum albumin and immunoglobulin G. Diffusivity of lysozyme did not follow the trend of decreasing diffusivity with increasing molecular mass, as observed for the rest of the proteins. In general, mass transfer coefficients were smaller compared to ion-exchange chromatography. Dynamic binding capacities for the model protein bovine serum albumin varied within a broad range. However, sorbents based on polymethacrylate showed a lower dynamic capacity than media based on Sepharose. Some sorbents could be clustered regarding binding capacity affected by salt. These sorbents exhibited a disproportional increase of binding capacity with increasing ammonium sulfate concentration. Recovery of proteins above 75% could be observed for all sorbents. Several sorbents showed a recovery close to 100%.

Currently, the selection of a hydrophobic interaction chromatography (HIC) sorbent for protein separation purposes is entirely based on empirical means. An attempt was made to characterize different HIC sorbents from various manufacturers. The selectivity was determined by isocratic pulse experiments of a set of reference proteins and an algorithm was developed to classify the sorbents according to their selectivity and hydrophobicity. The obtained semi-quantitative parameters take into account the dependence of salt on adsorption. The sorbent characteristics evaluated with the model proteins were compared to the separation of a real feedstock. A good agreement was achieved between the developed evaluation procedure and the separation behaviour of the real feed stock.