Accounting for host cell protein behavior in anion-exchange chromatography.

Host cell proteins (HCP) are a problematic set of impurities in downstream processing (DSP) as they behave most similarly to the target protein during separation. Approaching DSP with the knowledge of HCP separation behavior would be beneficial for the production of high purity recombinant biologics. Therefore, this work was aimed at characterizing the separation behavior of complex mixtures of HCP during a commonly used method: anion-exchange chromatography (AEX). An additional goal was to evaluate the performance of a statistical methodology, based on the characterization data, as a tool for predicting protein separation behavior. Aqueous two-phase partitioning followed by two-dimensional electrophoresis provided data on the three physicochemical properties most commonly exploited during DSP for each HCP: pI (isoelectric point), molecular weight, and surface hydrophobicity. The protein separation behaviors of two alternative expression host extracts (corn germ and E. coli) were characterized. A multivariate random forest (MVRF) statistical methodology was then applied to the database of characterized proteins creating a tool for predicting the AEX behavior of a mixture of proteins. The accuracy of the MVRF method was determined by calculating a root mean squared error value for each database. This measure never exceeded a value of 0.045 (fraction of protein populating each of the multiple separation fractions) for AEX. © 2016 American Institute of Chemical Engineers Biotechnol. Prog., 32:1453-1463, 2016.

Parameters affecting enzyme-assisted aqueous extraction of extruded sunflower meal.

Microscopic observation of sunflower meal before and after extraction indicated that extensive cellular disruption was achieved by extrusion, but that unextracted oil remained sequestered as coalesced oil within the void spaces of disrupted cotyledon cells. A full factorial design experiment was defined to develop aqueous extraction processing (AEP) with and without enzymes to improve vegetable oil extraction yields of extruded sunflower meal. This experimental design studied the influence of four parameters, agitation, liquid/solid (L/S) ratio, and cellulase and protease addition, on extraction yield of lipid and protein. Agitation and addition of cellulases increased oil extraction yield, indicating that emulsification of oil and alteration of the geometry of the confining cellular matrix were important mechanisms for improving yields. Protease and liquid-solid ratio of the extraction mixture did not have significant effects, indicating key differences with previously established soy oil extraction mechanisms. Maximum yields attained for oil and protein extraction were 39% and 90%, respectively, with the aid of a surfactant.

Process integration for recovery of recombinant collagen type I α1 from corn seed.

Because of safety concerns and product consistency issues with the use of animal-derived collagen, several recombinant protein expression hosts have been considered for recombinant collagen corn seed. Full length, triple-helical, recombinant collagen (rCIα1) is expressed as a fusion with a foldon domain, which must later be removed. Here we have examined integration of purification and foldon removal by comparing advantages of removal before or after purification, using salt precipitation as the main purification step. Because expression levels in available maize lines are low, Pichia-produced recombinant collagens, both with and without foldon, were added to corn seed germ at the extraction step. Salt precipitation of an acidic corn seed extract yielded 100% of the collagen without foldon at >70% purity without the pepsin pretreatment. With pepsin pretreatment, yield was 94.0% with purity of 76.5%. Analysis of the protein molecular weight distribution of the pre- and post-treatment extracts showed that the corn proteins are largely resistant to pepsin proteolysis, explaining why little benefit was obtained by pepsin treatment. In the absence of pepsin treatment, the recovery of rCIα1 with foldon was still above 90% but the purity was only 44%. This still represented at about 13-fold purification with a 2.7-fold volume reduction which would reduce the pepsin requirement for post-recovery foldon cleavage.

Proteomics-based, multivariate random forest method for prediction of protein separation behavior during cation-exchange chromatography.

The most significant cost of recombinant protein production lies in the optimization of the downstream purification methods, mainly due to a lack of knowledge of the separation behavior of the host cell proteins (HCP). To reduce the effort required for purification process development, this work was aimed at modeling the separation behavior of a complex mixture of proteins in cation-exchange chromatography (CEX). With the emergence of molecular pharming as a viable option for the production of recombinant pharmaceutical proteins, the HCP mixture chosen was an extract of corn germ. Aqueous two phase system (ATPS) partitioning followed by two-dimensional electrophoresis (2DE) provided data on isoelectric point, molecular weight and surface hydrophobicity of the extract and step-elution fractions. A multivariate random forest (MVRF) method was then developed using the three characterization variables to predict the elution pattern of individual corn HCP. The MVRF method achieved an average root mean squared error (RMSE) value of 0.0406 (fraction of protein eluted in each CEX elution step) for all the proteins that were characterized, providing evidence for the effectiveness of both the characterization method and the analysis approach for protein purification applications.

Utilizing protein-lean coproducts from corn containing recombinant pharmaceutical proteins for ethanol production.

Protein-lean fractions of corn (maize) containing recombinant (r) pharmaceutical proteins were evaluated as a potential feedstock to produce fuel ethanol. The levels of residual r-proteins in the coproduct, distillers dry grains with solubles (DDGS), were determined. Transgenic corn lines containing recombinant green fluorescence protein (r-GFP) and a recombinant subunit vaccine of Escherichia coli enterotoxin (r-LTB), primarily expressed in endosperm, and another two corn lines containing recombinant human collagen (r-CIα1) and r-GFP, primarily expressed in germ, were used as model systems. The kernels were either ground and used for fermentation or dry fractionated to recover germ-rich fractions prior to grinding for fermentation. The finished beers of whole ground kernels and r-protein-spent endosperm solids contained 127-139 and 138-155 g/L ethanol concentrations, respectively. The ethanol levels did not differ among transgenic and normal corn feedstocks, indicating the residual r-proteins did not negatively affect ethanol production. r-Protein extraction and germ removal also did not negatively affect fermentation of the remaining mass. Most r-proteins were inactivated during the mashing process used to prepare corn for fermentation. No functionally active r-GFP or r-LTB proteins were found after fermentation of the r-protein-spent solids; however, a small quantity of residual r-CIα1 was detected in DDGS, indicating that the safety of DDGS produced from transgenic grain for r-protein production needs to be evaluated for each event. Protease treatment during fermentation completely hydrolyzed the residual r-CIα1, and no residual r-proteins were detectable in DDGS.

Recovering corn germ enriched in recombinant protein by wet-fractionation.

Corn wet-fractionation processes (quick-germ fractionation and traditional wet milling) were evaluated as means of recovering fractions rich in recombinant collagen-related proteins that were targeted for expression in the germ (embryo) of transgenic corn. Transgenic corn lines accumulating a recombinant full-length human collagen type-I-alpha-1 (full-length rCIalpha1) or a 44-kDa rCIalpha1 fragment targeted for seed expression with an embryo-specific promoter were used. Factors to consider in efficient recovery processes are the distribution of the peptides among botanical parts and process recovery efficiency. Both recombinant proteins were distributed 62-64% in germ comprising about 8.6% of the dry grain mass; 34-38% in the endosperm comprising 84% of the dry grain mass; 1.7% in the pericarp comprising about 5% of the dry mass; and 1% in the tip-cap comprising 1.5-2% of the dry mass. The quick-germ method employed a short steeping period either in water or SO(2)-lactic acid solution followed by wet-milling degermination to recover a germ-rich fraction. Of the total recombinant protein expressed in germ, the quick-germ process recovered 40-43% of the total recombinant protein within 6-8% of the corn mass. The traditional corn wet-milling process produced higher purity germ but with lower recovery (24-26%) of the recombinant protein. The two quick-germ methods, using water alone or SO(2)-lactic acid steeping, did not substantially differ in rCIalpha1 recovery, and the quick-germ processes recovered germ with less leaching and proteolytic losses of the recombinant proteins than did traditional wet milling. Thus, grain fractionation enriched the recombinant proteins 6-fold higher than that of unfractionated kernels. Such enrichment may improve downstream processing efficiency and enable utilizing the protein-lean co-products to produce biofuels and biorenewable chemicals by fermenting the remaining starch-rich fractions.

Protein recovery from enzyme-assisted aqueous extraction of soybean.

Enzyme-assisted aqueous oil extraction from soybean is a "green" alternative to hexane extraction that must realize potential revenues from a value-added protein co-product. Three technologies were investigated to recover protein from the skim fraction of an aqueous extraction process. Ultrafiltration achieved overall protein yields between 60% and 64%, with solids protein content of 70%, and was effective in reducing stachyose content, with fluxes between 4 and 10 L/m(2) hr. Protein content was limited because of high retention of lipids and the loss of polypeptides below 13.6 kDa. Isoelectric precipitation was effective in recovering the minimally hydrolyzed proteins of skim, with a protein content of 70%, again limited by lipid content. However, protein recovery was only 30% because of the greater solubility of the hydrolyzed proteins. Recovery by the alternative of protein capture on dextran-grafted agarose quaternary-amine expanded bed adsorption resins decreased with decreasing polypeptide molecular weight. Proteins with molecular mass greater than 30 kDa exhibited slow adsorption rates. Expanded bed adsorption was most effective for recovery of proteins with molecular weight between 30 and 12 kDa. Overall, adsorption protein yields were between 14% and 17%.

Characterization of green-tissue protein extract from alfalfa (Medicago sativa) exploiting a 3-D technique.

There is a growing interest of pharmaceutical companies for plant-based production systems. To facilitate the general acceptance of plants as bioreactors, the establishment of efficient downstream operations is critical. It has been proposed that a better understanding of the properties of the contaminant proteins can benefit downstream processing design and operation. The coupled application of 2-DE with aqueous two-phase partitioning has been suggested as a practical 3-D method to characterize potential contaminant proteins from plant extracts. The application of this novel 3-D approach to a complex protein extract from alfalfa (Medicago sativa) containing a model recombinant protein (human granulocyte colony stimulating factor (hG-CSF)) resulted in the quantification of 55 protein spots. The 3-D properties (M(r), pI, and K(p)) obtained for 17 proteins comprising 69% of the alfalfa proteins, allowed the proposal of a prefractionation step as well as the identification of the target molecule (rG-CSF) from bulk of alfalfa proteins. The information obtained from this experimental approach was useful for the identification of the potential contaminant proteins that will occur in alfalfa when this plant is used as a host for recombinant proteins. Additionally, this method will assist in the design of adequate purification strategies for recombinant proteins expressed in alfalfa green tissue.

Purification and characterization of a transgenic corn grain-derived recombinant collagen type I alpha 1.

Corn offers advantages as a transgenic host for producing recombinant proteins required at large volumes (1,000's of tons per year) and low cost (less than US$50/kg) by generating them as co-products of biorefining. We describe the purification and characterization of a corn grain-derived mammalian structural protein having such market characteristics: a full length recombinant collagen type I alpha 1 (rCI alpha 1) chain. Material properties of interest are gelation behavior, which would depend on as yet unverified ability of corn to carry out post-translational prolyl hydroxylation and formation of triple helical conformation. The starting material was grain where the expression of rCI alpha 1 had been directed by an embryo-specific promoter. Purification consisted of extraction at low pH followed by membrane and chromatographic steps to isolate rCI alpha 1 for characterization. The amino acid composition and immunoreactivity of CI alpha 1 was similar to that of an analogous native human CI alpha 1 and to rCI alpha 1 produced by the yeast Pichia pastoris. Tandem mass spectrometry confirmed the primary sequence of the corn-derived rCI alpha 1 with 46% coverage. Fragments of the rCI alpha 1 chains were also observed, possibly caused by endogenous plant proteases. The corn-derived rCI alpha 1 had a low level of prolyl hydroxylation (approximately 1% versus 11%) relative to animal-derived CI alpha 1 and folded into its characteristic triple-helical structure as indicated by its resistance to pepsin digestion below its melting temperature of 26(o)C. The 29 amino acid foldon fused to the C-terminus to initiate triple helix formation was not cleaved from the rCI alpha 1 chains, but could be removed by pepsin treatment.

Fractionation of transgenic corn seed by dry and wet milling to recover recombinant collagen-related proteins.

Corn continues to be considered an attractive transgenic host for producing recombinant therapeutic and industrial proteins because of its potential for producing recombinant proteins at large volume and low cost as coproducts of corn seed-based biorefining. Efforts to reduce production costs have been primarily devoted to increasing accumulation level, optimizing protein extraction conditions, and simplifying the purification. In the present work, we evaluated two grain fractionation methods, dry milling and wet milling, to enrich two recombinant collagen-related proteins; thereby, reducing the amount and type of corn-derived impurities in subsequent protein extraction and purification steps. The two proteins were a full-length human recombinant collagen type I alpha 1(rCIalpha1) chain with telopeptides and peptide foldon to effect triple helix formation and a 44-kDa rCIalpha1 fragment. For each, approximately 60% of the rCIalpha1s in the seed was recovered in the dry-milled germ-rich fractions making up ca. 25% of the total kernel mass. For wet milling, approximately 60% of each was recovered in three fractions accounting for 20-25% of the total kernel mass. The rCIalpha1s in the dry-milled germ-rich fractions were enriched three to six times compared with the whole corn kernel, whereas the rCIalpha1s were enriched 4-10 times in selected wet-milled fractions. The recovered starch from wet milling was almost free of rCIalpha1. Therefore, it was possible to generate rCIalpha1-enriched fractions by both dry and wet milling along with rCIalpha1-free starch using wet milling. Because of its simplicity, the dry milling procedure could be accomplished on-farm thus minimizing the risk of inadvertent release of viable transgenic seeds.

Purification and characterization of a 44-kDa recombinant collagen I alpha 1 fragment from corn grain.

This paper demonstrates that a fibrous, repetitive amino acid sequence collagen-related protein, a 44-kDa fragment of human collagen I alpha 1 (CIalpha1), was expressed in corn grain molecularly equivalent to that produced in recombinant yeast. The recombinant CIalpha1 was extracted and purified from early generation plants having low levels of recombinant protein accumulation. It was selectively extracted at low pH and purified by ion exchange and gel filtration chromatography, resulting in a 44-kDa CIalpha1 with >70% purity and 60% recovery. The N-terminal sequence, amino acid composition, and immunoreactivity closely matched those of an analogous 44-kDa CIalpha1 fragment produced by the yeast Pichia . The corn-derived 44-kDa CIalpha1 had an intact protein mass of 44088 Da, which is within 0.2% of the mass calculated from the expected sequence. Tandem mass spectrometry confirmed the primary sequence with 78% coverage. The amino acid composition analysis indicated a low level of prolyl hydroxylation. Glycoprotein staining revealed no glycosylation.

Predicting protein retention time in ion-exchange chromatography based on three-dimensional protein characterization.

The isoelectric point (pI), molecular weight (M(W)) and aqueous two-phase partitioning coefficients of a set of model proteins were related to retention time in cation-exchange chromatography using partial least squares regression. A three-dimensional method which combined hydrophobic partitioning and two-dimensional electrophoresis was used to determine those three properties for a mixture of proteins. The regression models fit well (R(2)=0.913 and 0.873 for two resin types) considering the limited property basis, and were able to predict results for a small test set of proteins. The models showed that greater size and charge increased retention time, while the net influence of hydrophobicity depended on the base matrix type. This establishes the potential for the intended application to complex mixtures of host cell proteins.

Extraction of protein from distiller's grain.

We have investigated the feasibility of extracting the oil and protein from distiller's grain (DG) to obtain a higher-valued protein-rich product and a carbohydrate-rich residue better suited for conversion to fermentable sugars. Protein extractions based on aqueous ethanol, alkaline-ethanol, and aqueous enzyme treatments were compared. Three of the methods extracted a significant amount of the protein from dried, defatted DG (DDDG). The enzymatic extraction decreased the crude protein content in the solid phase for both milled and unmilled DDDG from 41% (dry weight) to approximately 10% (dry weight) protein in the residual solid; this corresponded to extraction of 90% of the protein in the original DDDG. The alkaline-ethanol extraction was similarly effective for milled but not unmilled DG. Simple extraction with alcohol was not as effective. Amino acid analysis of each protein extract was consistent with it consisting mainly of zeins. For the protease-assisted extractions, 95% the proteins were in the form of peptides smaller than 10kDa.

Purification of recombinant aprotinin from transgenic corn germ fraction using ion exchange and hydrophobic interaction chromatography.

Plants have attracted interest as hosts for protein expression because of the promise of a large production capacity and a low production cost. However, recovery costs remain a challenge as illustrated for recovery of recombinant aprotinin, a trypsin inhibitor, with removal of native corn trypsin inhibitor from transgenic corn (Azzoni et al. in Biotechnol Bioeng 80:268-276, 2002). When expression is targeted to corn grain fractions, dry milling can separate germ and endosperm fractions. Hence, only the product-containing fraction needs to be extracted, reducing the cost of extraction and the impurity level of the extract. Selective extraction conditions can reduce impurity levels to the point that low-cost adsorbents can result in relatively high purity levels. In this work, we attempted to achieve comparable purity with these lower cost methods. We replaced whole grain extraction and purification of recombinant aprotinin with sequential trypsin affinity and IMAC steps with an alternative of germ fraction extraction and purification with ion exchange and hydrophobic interaction chromatography (HIC). Using germ extraction at acidic pH supplemented with heat precipitation to remove additional host proteins resulted in a higher specific activity feed to the chromatographic steps. The cation exchange step provided 7.6x purification with 76.4% yield and no sodium dodecyl sulfate-polyacrylamide gel electrophoresis detectable native corn trypsin inhibitor. After the HIC step (2.7x step purification with 44.0% yield), the final product had a specific activity that was 75.3% of that of the affinity-purified aprotinin.

A method for three-dimensional protein characterization and its application to a complex plant (corn) extract.

Knowledge of host protein properties is critical for developing purification methods for recombinant proteins from a specific host, or for choosing suitable hosts and targeted expression tissues for a specific recombinant protein. A method to obtain a three-dimensional (3D) map (surface hydrophobicity (SH), isoelectric point (pI), and molecular weight (MW)), of a host's aqueous soluble protein properties was developed. The method consists of hydrophobic partitioning in a PEG 3350 (15.7%)-Na(2)SO(4) (8.9%)-NaCl (3%) aqueous two-phase (ATP) system followed by quantitative, 2D-electrophoretic characterization of the proteins of each equilibrium phase and the original extract. The pI and MW of host proteins were obtained directly through 2D electrophoresis. The partition coefficients of individual proteins were obtained by quantitative matching of protein spots in the top and bottom phase gels and calculating the protein partition coefficients from this information. Correlation of the partition coefficient to a SH scale was established by partitioning several model proteins with known surface hydrophobicities in the same ATP system. The inclusion of the extract gel provided for a spot selection criterion based on satisfactory mass balance closure. The method is illustrated by application to a mixture of model proteins and to complex mixtures, that is, corn germ proteins extracted at pH 7 and pH 4.

Aqueous two-phase extraction for protein recovery from corn extracts.

Corn has been used as an expression host for several recombinant proteins with potential for large-scale production. Cost-effective downstream initial recovery, separation and concentration remain a challenge. Aqueous two-phase (ATP) partitioning has been used to recover and concentrate proteins from fermentation broths and offers advantages for integration of those steps with biomass removal. To examine the applicability of ATP partitioning to recombinant protein purification from corn endosperm and germ, ATP system parameters including poly(ethylene glycol) (PEG) molecular weight (MW), phase-forming salt, tie line length (TLL), and pH were manipulated to control partitioning of extracted native proteins from each fraction. Moderate PEG MW, reduction of phase ratio, and added NaCl effected complete recovery of the hydrophobic model protein lysozyme in the top phase with ca. 5x enrichment and illustrates a favorable match of recombinant protein characteristics, expression host, and separation method. Furthermore, integration of protein extraction with the partitioning reduced the load of contaminating host proteins relative to the more traditional separate steps of extraction followed by partitioning. Performance of the integrated partitioning was hindered by endosperm solids loading, whereas for germ, which has ca. 35x higher aqueous soluble protein, the limit was protein solubility. For more hydrophilic model proteins (the model being cytochrome c), effective separation required further reduction of PEG MW to effect more partitioning of host proteins to the top phase and enrichment of the model protein in the lower phase. The combination of PEG MW of 1450 with 8.5 wt.% NaCl addition (Na(2)SO(4) as the phase-forming salt) provided for complete recovery of cytochrome c in the lower phase with enrichment of 9x (germ) and 5x (endosperm). As a result of lower-phase recovery, the advantage of simultaneous removal of solids is lost. The lower solubility of native endosperm proteins results in higher purity for the same enrichment.

Extraction of recombinant dog gastric lipase from transgenic corn seed.

Several approaches were examined for extracting the relatively hydrophobic protein recombinant dog gastric lipase (rDGL) expressed in the endosperm of transgenic corn seed. The first approach used minimal processing of the seed before extraction (i.e. simple grinding of whole seed) followed by selective extraction to eliminate 72% of contaminant proteins without compromising rDGL recovery from the meal of whole grain. The second approach added defatting of the whole grain meal to reduce the amount of detergent in the subsequent step for extracting rDGL. The third approach incorporated dry-milling of the corn to recover an endosperm rich fraction, followed by extraction of this fraction. The dry milling strategy was most effective, resulting in recovery of 35 U rDGL/g of corn seed (50 U/g of endosperm) with a specific activity of 9 U/mg compared to 22 U and 3 U/mg for the first strategy and 36 U and 3.7 U/mg for the second. The reductions in host protein contamination and lower detergent levels of the endosperm route should simplify downstream purification steps.

Enzymatic assay method for evaluating the lipase activity in complex extracts from transgenic corn seed.

A colorimetric method was established to determine the activity of recombinant lipase in extracts from transgenic corn seed. The system was an oil-in-water emulsion that was stabilized by a surfactant to accommodate the organic phase substrate and aqueous phase enzyme. The lipase activity was measured by monitoring the release of nitrophenol at 346 nm from the substrate, 4-nitrophenyl butyrate. Emulsions prepared with various surfactant types and concentrations were tested. For each surfactant, the measured activity was greatest when the surfactant concentration was close to the critical micelle concentration, consistent with the changing trend of oil droplet size as a function of surfactant concentration. The optimal system, with 0.01% (w/w) Tween 80, demonstrated good reproducibility, high sensitivity, robustness, and a linear response to lipase concentration.

Problem-based learning biotechnology courses in chemical engineering.

We have developed a series of upper undergraduate/graduate lecture and laboratory courses on biotechnological topics to supplement existing biochemical engineering, bioseparations, and biomedical engineering lecture courses. The laboratory courses are based on problem-based learning techniques, featuring two- and three-person teams, journaling, and performance rubrics for guidance and assessment. Participants initially have found them to be difficult, since they had little experience with problem-based learning. To increase enrollment, we are combining the laboratory courses into 2-credit groupings and allowing students to substitute one of them for the second of our 2-credit chemical engineering unit operations laboratory courses.

Applicability of the stoichiometric displacement model to description of the retention behavior of charged-fusion proteins during fast protein liquid chromatography.

The applicability of the stoichiometric displacement model (SDM) to description of the retention behavior of charged-fusion proteins in large ion exchange resin (approximately 90 microm diameter) packed column was studied. Proteins were characterized by SDM for isocratic elution. The parameters were subsequently used to evaluate their suitability in predicting protein retention and peak width under gradient elution. The proteins were beta-glucuronidase (GUS) and its fusions with polypeptides of 5, 10 and 15 aspartic acids at the C-terminal of the wild-type GUS. Predictions of retention time were within 10% of the experiment results. The plate number obtained at high salt concentration from isocratic elution was used as a first estimate for predictions of peak width. The results show that the SDM is sufficient to describe the binding equilibrium of fusion proteins in ion-exchange columns packed with large resin particles. In addition, the binding mechanism between fusion proteins and the ion exchanger is explored with the assistance of comparative molecular modeling.