Recovery of proteins from corn and soybean extracts by membrane adsorption.

Efficient separation strategies for the recovery of high-value proteins (native or recombinant) from plant agriculture are an important aspect of many different processes, from biopharmaceuticals to byproduct recovery during biofuel production. Here we report the use of membrane adsorption for the recovery of proteins from soybean and corn extracts, and compare the results with packed bed adsorption. Two alternative operating modes were investigated, a flowthrough strategy and a bind and elute method. Overall, membrane adsorption provided faster throughput, and had equal or slightly higher dynamic binding capacities compared with resin beads, without compromising yield and purity of the chosen target. Soybean was found to be an ideal plant host when capturing native protein on an anion exchange medium. This provided an opportunity to capture a large percentage (>80%) of native protein as the product, and/or allowed for elevated enrichment factors (>20) during purification of a recombinant target with pI > 7.0, using a flowthrough approach. On the other hand, for corn, a single ion-exchange step was not able to capture more than 60% of the native protein. However, the bind and elute method with corn as the host for a recombinant product allowed for higher enrichment factors compared to soybean. In all cases, the concentration of a recombinant protein (as dictated by expression level) was found to play a significant role in the level of dynamic binding capacity, with higher concentration leading to elevated capacity. Likewise, a higher concentration of competing proteins was shown to decrease the overall capacity of a recombinant target.

Antibody capture from corn endosperm extracts by packed bed and expanded bed adsorption.

Topical treatments of chronic infections with monoclonal antibodies will require large quantities of antibodies. Because plants have been proven capable of producing multisubunit antibodies and provide for large-scale production, they are likely hosts to enable such applications. Recovery costs must also be low because of the relatively high dosages required. Hence, we have examined the purification of a human secretory antibody from corn endosperm extracts by processing alternatives of packed bed and expanded bed adsorption (EBA). Because of the limited availability of the transgenic corn host, the system was modeled by adding the antibody to extracts of nontransgenic corn endosperm. Complete clarification of a crude extract followed by packed bed adsorption provided antibody product in 75% yield with 2.3-fold purification (with antibody accounting for 24% of total protein). The small size of the packed bed, cation-exchange resin SP-Sepharose FF and the absence of a dense core (present in EBA resins) allowed for more favorable breakthrough performance compared to EBA resins evaluated. Four adsorbents specifically designed for EBA operation, with different physical properties (size and density), chemical properties (ligand), and base matrices were tested: SP-steel core resin (UpFront Chromatography), Streamline SP and Streamline DEAE (Amersham Biosciences), and CM Hyper-Z (BioSepra/Ciphergen Biosystems). Of these, the small hyperdiffuse-style resin from BioSepra had the most favorable adsorption characteristics. However, it could not be utilized with crude feeds due to severe interactions with corn endosperm solids that led to bed collapse. UpFront SP-steel core resin, because of its relatively smaller size and hence lower internal mass transfer resistance, was superior to the Streamline resins and operated successfully with application of a crude corn extract filtered to remove all solids of >44 microm. However, the EBA performance with this adsorbent provided a yield of only 61% and purification factor of 2.1 (with antibody being 22% of total protein). Process simulation showed that capital costs were roughly equal between packed and expanded bed processes, but the EBA design required four times greater operating expenditures. The use of corn endosperm as the starting tissue proved advantageous as the amount of contaminating protein was reduced approximately 80 times compared to corn germ and approximately 600 times compared to canola. Finally, three different inlet designs (mesh, glass beads, and mechanical mixing) were evaluated on the basis of their ability to produce efficient flow distribution as measured by residence time distribution analysis. All three provided adequate distribution (axial mixing was not as limiting as mass transfer to the adsorption process), while resins with different physical properties did not influence flow distribution efficiency values (i.e., Peclet number and HETP) when operated with the same inlet design.

Compatibility of column inlet and adsorbent designs for processing of corn endosperm extract by expanded bed adsorption.

Corn has emerged as a viable host for expression of recombinant proteins; targeted expression to the endosperm has received particular attention. The protein extracts from corn endosperm differ from those of traditional hosts in regard to the nature of residual solids and extracted matrix contaminants. Each of these differences presents reasons for considering expanded bed adsorption for product capture and new considerations for limitations of the method. In this work three inlet-flow distribution devices (mesh, glass ballotini, and localized mixing) and six adsorbents with different physical (size and density), chemical (ligand), and base matrix properties were evaluated to determine conditions compatible with processing of crude corn endosperm extract by expanded bed adsorption. Of the inlet devices evaluated, the design with localized mixing at the inlet (as produced commercially by UpFront Chromatography A/S, Copenhagen, DK) allowed solids up to 550 microm into the column without clogging for all flow rates evaluated. A mesh at the inlet with size restriction of either 50 microm or 80 microm became clogged with very small corn particles (< 44 microm). When glass ballotini was used, large particles (550 microm) passed through for high flow rates (570 cm/h), but even small (< 44 microm) particles became trapped at a lower flow rate (180 cm/h). The physical and chemical properties of the resin determined whether solids could be eluted. The denser UpFront adsorbents allowed for complete elution of larger and more concentrated corn solids than the currently available Amersham Streamline adsorbents (Amersham Biosciences, Piscataway, NJ) as a result of the former's higher flow rate for the desired 2x expansion (570 cm/h for UpFront vs. 180 cm/h for Streamline). All corn solids < 162 microm eluted through nonderivatized UpFront resin. Larger corn solids began to accumulate due to their elevated sedimentation velocities. Feeds of < 44 microm solids at 0.45% and 2.0% dry weight successfully eluted through ion exchange adsorbents (DEAE and SP) from UpFront. However, significant accumulation occurred when the solids size increased to a feed of < 96 microm solids, thus indicating a weak interaction between corn solids and both forms of ion exchange ligands. Expanded beds operated with Streamline ion exchange adsorbents (DEAE and SP) did not allow full elution of corn solids of < 44 microm. A hyperdiffuse style EBA resin produced by Biosepra (Ciphergen Biosystems, Fremont, CA) with CM functionality showed a severe interaction with corn solids that collapsed the expanded bed and could not be eliminated with elevated flow rates or higher salt concentration.

Host selection as a downstream strategy: polyelectrolyte precipitation of beta-glucuronidase from plant extracts.

Host selection can be a strategy to simplify downstream processing for protein recovery. Advancing capabilities for using plants as hosts offers new host opportunities that have received only limited attention from a downstream processing perspective. Here, we investigated the potential of using a polycationic precipitating agent (polyethylenimine; PEI) to precipitate an acidic model protein (beta-glucuronidase; GUS) from aqueous plant extracts. To assess the potential of host selection to enhance the ease of recovery, the same procedure was applied to oilseed extracts of canola, corn (germ), and soy. For comparison, PEI precipitation of GUS was also evaluated from a crude bacterial fermentation broth. Two versions of the target protein were investigated--the wild-type enzyme (WTGUS) and a genetically engineered version containing 10 additional aspartates on each of the enzyme's four homologous subunits (GUSD10). It was found that canola was the most compatible expression host for use with this purification technique. GUS was completely precipitated from canola with the lowest dosage of PEI (30 mg PEI/g total protein), and over 80% of the initial WTGUS activity was recovered with 18-fold purification. Precipitation from soy gave yields over 90% for WTGUS but only 1.3-fold enrichment. Corn, although requiring the most PEI relative to total protein to precipitate (210 mg PEI/g total protein for 100% precipitation), gave intermediate results, with 81% recovery of WTGUS activity and a purification factor of 2.6. The addition of aspartate residues to the target protein did not enhance the selectivity of PEI precipitation in any of the systems tested. In fact, the additional charge reduced the ability to recover GUSD10 from the precipitate, resulting in lower yields and enrichment ratios compared to WTGUS. Compared to the bacterial host, plant systems provided lower polymer dosage requirements, higher yields of recoverable activity and greater purification factors.