Search

In this work, the interfacing of a poly(ethylene glycol) (PEG)-phosphate aqueous two-phase system with hydrophobic interaction chromatography (HIC) for primary recovery of an intracellular protein was evaluated. As a model protein, a recombinant cutinase furnished with a tryptophan-proline (WP) peptide tag was used and produced intracellularly in Escherichia coli (E. coli). E. coli cell homogenate was partitioned in a two-phase system and the top phase yield, concentration and purity of the tagged ZZ-cutinase-(WP)4 was evaluated as function of polymer sizes, system pH and phase volume ratio. The partition behaviour of cell debris, total protein and endotoxin was also monitored. In the HIC part, the chromatographic yield and purity was investigated with respect to ligand hydrophobicity, dilution of loaded top phase and elution conditions. Based on the results, a recovery process was demonstrated where a PEG 1500-K-Na phosphate salt aqueous two-phase system was interfaced with a HIC column. The interfacing was facilitated by the Trp-tagged peptide. The tagged ZZ-cutinase-(WP)4 was obtained in a PEG-free phase and purified to >95% purity according to silver stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels with a total yield of 83% during the two-step recovery process.

Endoglucanases (EGI) (endo-1,4-beta-D-glucan-4-glucanohydrolase, EC 3.2.1.4, Ce17B) of Trichoderma reesei are industrially important enzymes. Thus, there is a great need for development of a primary recovery method suitable for large-scale utilization. In this study we present a concept applicable for large-scale purification of an EGI fusion protein by one-step extraction in a poly(ethylene glycol) PEG-sodium/potassium phosphate aqueous two-phase system. EGI is a two-module enzyme composed of an N-terminal catalytic module and a C-terminal cellulose binding module (CBM) separated by a glycosylated linker region. Partitioning of six different EGI constructs, containing the C-terminal extensions (WP)2, (WP)4 or the amphiphilic protein hydrophobin I (HFB) of T. reesei instead of the CBM were studied to evaluate if any of the fusions could improve the partition coefficient sufficiently to be suitable for large-scale production. All constructs showed improved partitioning in comparison to full length EGI. The (WP)4 extensions resulted in 26- to 60-fold improvement of partition coefficient. Consequently, a relative minor change in amino acid sequence on the two-module protein EGI improved the partition coefficient significantly in the PEG 4000-sodium/potassium phosphate system. The addition of HFBI to EGI clearly enhanced the partition coefficient (K=1.2) in comparison to full-length EGI (K=0.035). Partitioning of the construct with (WP)4 fused to the catalytic module and a short sequence of the linker [EGI(core-P5)(WP)4] resulted in the highest partition coefficient (K=54) and a yield of 98% in the PEG phase. Gel electrophoresis showed that the construct with the (WP)4 tag attached after a penta-proline linker could be purified from the other bulk proteins by only a single-step separation in the PEG 4000-sodium/potassium phosphate system. This is a major improvement in comparison with the previously studied model (ethylene oxide-propylene oxide)-dextran system. Hence, this construct will be suitable for further optimization of the extraction of the enzyme in a PEG 4000-sodium/potassium phosphate system from culture filtrate.

A method to quantify the impact of proteolysis on accumulation of recombinant proteins in E. coli is described. A much smaller intracellular concentration of staphylococcal protein A (SpA) (14.7 mg. g(-1)) compared to the fusion protein SpA-betagalactosidase (138 mg. g(-1)) is explained by a very high proteolysis rate constant of SpA. The SpA synthesis rate reached a maximum one hour after induction and gradually decreased to half of this value at the end of the cultivation. The decrease of the synthesis rate and the 1st order kinetics of proteolysis lead to an equilibrium between synthesis and degradation of SpA from 2 h after induction. This resulted in no further SpA accumulation in cells, though synthesis continued for at least 10 h. Similar experiments with recombinant protein ZZT2 also revealed that most of the synthesized product was degraded. The order of proteolysis kinetics depended on the concentration of the recombinant protein: at low concentrations both SpA and ZZT2 were degraded according to first order kinetics, while at high concentrations ZZT2 was degraded according to zero order kinetics. In a protease Clp mutant the degradation rate decreased and intracellular concentration of ZZT2 increased from 50 mg. g(-1) to 120 mg. g(-1). The measurements of proteolysis rate throughout the cultivation enabled calculation of a hypothetical accumulation of the product assuming complete stabilization. In this case the concentration would have increased from 50 to 280 mg. g(-1) in 11 h. Thus, this method reveals the potential to increase the productivity by eliminating proteolysis.

A series of proteins and one membrane-bound peptide have been partitioned in aqueous two-phase systems consisting of micelle-forming block copolymers from the family of Pluronic block copolymers as one polymer component and dextran T500 as the other component. The Pluronic molecule is a triblock copolymer of the type PEO-PPO-PEO, where PEO and PPO are poly(ethylene oxide) and poly(propylene oxide), respectively. Two different Pluronic copolymers were used, P105 and F68, and the phase diagrams were determined at 30 degrees C for these polymer systems. Since the temperature is an important parameter in Pluronic systems (the block copolymers form micellar-like aggregates at higher temperatures) the partitioning experiments were performed at 5 and 30 degrees C, to explore the effect of temperature-triggered micellization on the partitioning behaviour. The temperatures correspond to the unimeric (single Pluronic chain) and the micellar states of the P105 polymer at the concentrations used. The degree of micellization in the F68 system was lower than that in the P105 system, as revealed by the phase behaviour. A membrane-bound peptide, gramicidin D, and five different proteins were partitioned in the above systems. The proteins were lysozyme, bovine serum albumin, cytochrome c, bacteriorhodopsin and the engineered B domain of staphylococcal protein A, named Z. The Z domain was modified with tryptophan-rich peptide chains in the C-terminal end. It was found that effects of salt dominated over the temperature effect for the water-soluble proteins lysozyme, bovine serum albumin and cytochrome c. A strong temperature effect was observed in the partitioning of the integral membrane protein bacteriorhodopsin, where partitioning towards the more hydrophobic Pluronic phase was higher at 30 degrees C than at 5 degrees C. The membrane-bound peptide gramicidin D partitioned exclusively to the Pluronic phase at both temperatures. The following trends were observed in the partitioning of the Z protein. (i) At the higher temperature, insertion of tryptophan-rich peptides increased the partitioning to the Pluronic phase. (ii) At the lower temperature, lower values of K were observed for ZT2 than for ZT1.

Genetic engineering has been used to construct fusion proteins with tryptophan containing peptides. The peptides and the fusion proteins have been partitioned in aqueous two-phase systems of poly(ethylene glycol) (PEG)-dextran and Ucon-dextran. The studied model protein was ZZT0, where Z is an engineered domain of domain B of staphylococcal protein A. The specially designed hydrophobic peptides, Ala-Trp-Trp-Pro (T1) and (Ala-Trp-Trp-Pro)2 (T2), have been inserted into ZZT0, to give the peptide-protein fusions ZZT1 and ZZT2. In the experimental studies it was found that T1 and T2 preferred the PEG phase and even more the Ucon phase over the dextran phase. For T2 the partitioning was more one sided than for T1. For the fusion proteins, ZZT1 and ZZT2, the partitioning was enhanced into the PEG or Ucon rich phase as compared to ZZT0. The effects were lower than expected from independent contributions to the partition coefficient from the protein and the peptides. A heterogeneous lattice model was used to calculate theoretical peptide and protein partition coefficients. The calculations could reproduce the qualitative features of the experimental data. The model results suggest that a part of these experimentally observed effects is due to a depletion zone, i.e. a zone of reduced polymer concentration around the protein. The experimental results indicate a further reduction of the partition coefficient, beyond that predicted by the lattice calculations. A possible folding of the inserted peptide is discussed as a plausible mechanism for this further reduction in the partition coefficient.

Two different tetrapeptides, AlaTrpTrpPro and AlaIleIlePro, were inserted near the C-terminus of the protein ZZT0. The Trp-rich peptide unit strongly increased both the partitioning of ZZT0 into the polyethylene glycol (PEG)-rich phase in a PEG-potassium phosphate aqueous two-phase system and its retention on PEG and propyl hydrophobic interaction chromatographic columns with potassium phosphate as eluent. Both the partitioning and the retention increased with increasing number of Trp-rich peptide units inserted into ZZT0. Insertion of Ile-rich tetrapeptide units affected the partitioning and retention to a much lesser extent. Partition data also indicated a folding of inserted Trp tetrapeptides units, probably to minimize their water contact.

Four different beta-galactosidase fusion proteins have been partitioned in poly(ethylene glycol) (PEG) 4000/potassium phosphate aqueous two-phase systems. The partition coefficients (K) of staphylococcal protein A-beta-galactosidase (SpA beta gal) (K = 3.5) and staphylococcal protein A-streptococcal protein G-beta-galactosidase (AG beta gal) (K = 2.8) were compared with the partition coefficients of their constituent molecules, beta-galactosidase, SpA, and protein AG. It was found that by fusing beta-galactosidase to the smaller proteins SpA and protein AG, their partition coefficients were increased four to five times. Experimental data were fitted into, and found to agree with, the Albertsson partition model of interacting molecules. The compatibility with PEG and potassium phosphate of beta-galactosidase, SpA, and two different versions of the SpA beta gal protein, displayed as precipitation curves, showed a relationship to the protein partition coefficients in PEG/potassium phosphate systems. High solubility in one phase component was accompanied by preferential partitioning to the phase rich in the same component in the PEG/potassium phosphate system. Also, a changed linker region in SpA beta gal resulted in a more soluble protein. This, together with the improved K values of the target proteins by fusion, shows that it is possible to use beta-galactosidase as an affinity handle.