Ultra scale-down device to predict dewatering levels of solids recovered in a continuous scroll decanter centrifuge.

During centrifugation operation, the major challenge in the recovery of extracellular proteins is the removal of the maximum liquid entrapped within the spaces between the settled solids-dewatering level. The ability of the scroll decanter centrifuge (SDC) to process continuously large amounts of feed material with high concentration of solids without the need for resuspension of feeds, and also to achieve relatively high dewatering, could be of great benefit for future use in the biopharmaceutical industry. However, for reliable prediction of dewatering in such a centrifuge, tests using the same kind of equipment at pilot-scale are required, which are time consuming and costly. To alleviate the need of pilot-scale trials, a novel USD device, with reduced amounts of feed (2 mL) and to be used in the laboratory, was developed to predict the dewatering levels of a SDC. To verify USD device, dewatering levels achieved were plotted against equivalent compression (Gtcomp ) and decanting (Gtdec ) times, obtained from scroll rates and feed flow rates operated at pilot-scale, respectively. The USD device was able to successfully match dewatering trends of the pilot-scale as a function of both Gtcomp and Gtdec , particularly for high cell density feeds, hence accounting for all key variables that influenced dewatering in a SDC. In addition, it accurately mimicked the maximum dewatering performance of the pilot-scale equipment. Therefore the USD device has the potential to be a useful tool at early stages of process development to gather performance data in the laboratory thus minimizing lengthy and costly runs with pilot-scale SDC.

Prediction and verification of centrifugal dewatering of P. pastoris fermentation cultures using an ultra scale-down approach.

Recent years have seen a dramatic rise in fermentation broth cell densities and a shift to extracellular product expression in microbial cells. As a result, dewatering characteristics during cell separation is of importance, as any liquor trapped in the sediment results in loss of product, and thus a decrease in product recovery. In this study, an ultra scale-down (USD) approach was developed to enable the rapid assessment of dewatering performance of pilot-scale centrifuges with intermittent solids discharge. The results were then verified at scale for two types of pilot-scale centrifuges: a tubular bowl equipment and a disk-stack centrifuge. Initial experiments showed that employing a laboratory-scale centrifugal mimic based on using a comparable feed concentration to that of the pilot-scale centrifuge, does not successfully predict the dewatering performance at scale (P-value <0.05). However, successful prediction of dewatering levels was achieved using the USD method (P-value ≥0.05), based on using a feed concentration at small-scale that mimicked the same height of solids as that in the pilot-scale centrifuge. Initial experiments used Baker's yeast feed suspensions followed by fresh Pichia pastoris fermentation cultures. This work presents a simple and novel USD approach to predict dewatering levels in two types of pilot-scale centrifuges using small quantities of feedstock (<50 mL). It is a useful tool to determine optimal conditions under which the pilot-scale centrifuge needs to be operated, reducing the need for repeated pilot-scale runs during early stages of process development.

Step change in the efficiency of centrifugation through cell engineering: co-expression of Staphylococcal nuclease to reduce the viscosity of the bioprocess feedstock.

Cell engineering to enable step change improvements in bioprocessing can be directed at targets other than increasing product titer. The physical properties of the process suspension such as viscosity, for example, have a major impact on various downstream processing unit operations. The release of chromosomal DNA during homogenization of Escherichia coli and its influence on viscosity is well-recognized. In this current article we demonstrate co-expression of Staphylococcus aureus nuclease in E. coli to reduce viscosity through auto-hydrolysis of nucleic acids. Viscosity reduction of up to 75% was achieved while the particle size distribution of cell debris was maintained approximately constant (d(50) = 0.5-0.6 microm). Critically, resultant step change improvements to the clarification performance under disc-stack centrifugation conditions are shown. The cell-engineered nuclease matched or exceeded the viscosity reduction performance seen with the addition of exogenous nuclease removing the expense and validation issues associated with such additions to a bioprocess. The resultant material dramatically altered performance in scale-down mimics of continuous disc-stack centrifugation. Laboratory scale data indicated that a fourfold reduction in the settling area of a disc-stack centrifuge can be expected due to a less viscous process stream achieved through nuclease co-expression with a recombinant protein.

A decision-support model for evaluating changes in biopharmaceutical manufacturing processes.

A simulation is described that evaluates the impacts of altering bio-manufacturing processes. Modifications designed to improve production levels, times and costs were assessed, including increasing feed volumes/titres, replacing initial downstream stages with packed or expanded bed affinity steps and removing ion exchange steps. Options were evaluated for manufactured product mass, COG, batch times and development costs and timescales. Metrics were combined using multi-attribute-decision-making techniques generating a single assessment metric for each option. The utility of this approach was illustrated by application to an FDA-approved process manufacturing rattlesnake anti-venom (Protherics U.K.). Currently, ovine serum containing anti-venom IgG is purified by precipitation/centrifugation, prior to antibody proteolysis by papain. An ion exchanger removes F(C), before affinity chromatography yields the final anti-venom. An expanded bed affinity column operating with an 80% higher IgG titre, 66% higher feed volume and without the ion exchanger delivered the best multi-attribute-decision-making value, potentially providing the most desirable alternative.

Ultra scale-down approach for the prediction of full-scale recovery of ovine polycolonal immunoglobulins used in the manufacture of snake venom-specific Fab fragment.

In this article, we describe a new approach that allows the prediction of the performance of a large-scale integrated process for the primary recovery of a therapeutic antibody from an analysis of the individual unit operations and their interactions in an ultra scale-down mimic of the process. The recovery process consisted of four distinct unit operations. Using the new approach we defined the important engineering parameters in each operation that impacted the overall recovery process and in each case verified its effect by a combination of modelling and experimentation. Immunoglobulins were precipitated from large volumes of dilute blood plasma and the precipitated flocs were recovered by centrifugal separation from the liquor containing contaminating proteins, including albumin. The fluid mechanical forces acting on the precipitate and the time of exposure to these forces were used to define a time-integrated fluid stress. This was used as a scaling factor to predict the properties of the precipitated flocs at large scale. In the case of centrifugation, the performance of a full-scale disc stack centrifuge was predicted. This was achieved from a computational fluid dynamics (CFD) analysis of the flow field in the centrifuge coupled with experimental data obtained from the precipitated immunoglobulin flocs using the scale-down precipitation tank, a rotating shear device, and a standard swing-out rotor centrifuge operating under defined conditions. In this way, the performance of the individual unit operations, and their linkage, was successfully analysed from a combination of modelling and experiments. These experiments required only millilitre quantities of the process material. The overall performance of the large-scale process was predicted by tracking the changes in physical and biological properties of the key components in the system, including the size distribution of the antibody precipitates and antibody activity through the individual unit operations in the ultra scale-down process flowsheet.

Ultra scaledown to predict filtering centrifugation of secreted antibody fragments from fungal broth.

Extracellularly expressed anti-hen egg lysozyme single-chain antibody fragments (scFv) produced by Aspergillus awamori were recovered using filtering centrifugation. Two filtering centrifuges with 0.5- and 30-L capacities were used to represent laboratory- and pilot-scale equipment, respectively. Critical regime analysis using the computational fluid dynamics (CFD) technique provided information about the local energy dissipation rates in both units. Experimental data indicated loss of scFv activity for energy dissipation rates above about 2.0 x 10(4) W kg(-1). This loss of activity increased in the presence of gas-liquid interfaces during filtering centrifugation. An ultra scaledown filtering centrifuge with a maximum working volume of 35 mL was designed to mimic the operating conditions identified by the critical regime analysis for the laboratory- and pilot-plant-scale units. The recovered scFv activity levels and the separation performance of the three units were comparable when operated at equal maximum energy dissipation rates.

A comparison of the process issues in expressing the same recombinant enzyme periplasmically in Escherichia coli and extracellularly in Streptomyces lividans.

The choice of a host for the production of a biological molecule will have a significant effect on isolation and purification procedures employed. This paper makes a comparison between the production of a single enzyme, a recombinant alpha-amylase, in Escherichia coli and Streptomyces lividans, on a small scale. It defines the differences in the cultivation and in the isolation stages and also describes the impact of the expression system on later downstream processing steps. At the cultivation stage, the specific productivity of the E. coli in units per gram per hour is four times that of the S. lividans while the total biomass yields are of the same order. The initial volume for downstream processing of S. lividans is six-fold larger and the total protein released into the extracellular medium is three times greater than E. coli, however, the recoverable yield from the E. coli is a fifth of that obtained from the S. lividans and requires three additional stages prior to chromatography. Even with these stages the final specific activity is 64% of the S. lividans. The results indicate the need to consider the whole process when making such comparisons.

Disruption of leaves and initial extraction of wildtype CPMV virus as a basis for producing vaccines from plants.

Wildtype cowpea mosaic virus (CPMV) was extracted from fresh and frozen plant material by methods suitable for large-scale application. Deep freezing, crushing, and thawing in water or buffers gave 0.6+/-0.2 mg g(-1) of virus after up to 24 h. Release from sliced fresh leaves was lower at 0.14+/-0.03 mg g(-1). Homogenisation of frozen leaves for 1 min increased yield to a maximum, on average of 3.5 mg g(-1) but varying between batches from 2.2 to 4.8 mg g(-1) virus Long term storage at -80 degrees C increased subsequent yield by 2 mg g(-1) per year on average; the maximum was 10.4+/-1.9 mg g(-1) (665 days storage). Within a batch, similar yields were obtained between individual fresh plants, and from frozen versus fresh leaves. After homogenisation for 1 min, 90% of debris particles were smaller than 12 microm, half under 5 microm and 10% less than 1 microm. Homogenate (4% dry weight) was rheologically complex, exhibiting shear thinning with hysteresis at low shear rates which bears on subsequent processing. At shear rates above 200 s(-1), its apparent viscosity was 0.02 N s m(-2).

Factors affecting the production of a single-chain antibody fragment by Aspergillus awamori in a stirred tank reactor.

A recombinant strain of Aspergillus awamori expressing anti-lysozyme single chain antibody fragments (scFv), under the control of a xylanase promoter, was studied in order to investigate the impact of medium, induction regime and protease production on the expression of the product. Experiments with the time of induction showed that the optimum results are achieved when induction is started in the late exponential phase (21 h after inoculation) improving the titer of the product from 14.5 mg L(-1), obtained in the early exponential phase (7 h after inoculation), to 16.2 mg L(-1). A 100% increase of the carbon (fructose) and nitrogen (ammonium sulfate) sources in the growth medium resulted in an increase in product concentration from 16.2 to 108.9 mg L(-1) and an increase in maximum dry cell weight from 7.5 to 11.5 g L(-1). A 50% reduction in the concentration of the inducer resulted in an increase in the product yield from 10 mg g(-1) dry cell weight to 12 mg g(-1). Proteolytic enzymes were produced during the fermentation up to concentrations equivalent to 1.4 g L(-1) trypsin, but they had no detrimental effect on the concentration of the antibody fragment.

Effects of growth medium selection on plasmid DNA production and initial processing steps.

Cultures of recombinant Escherichia coli containing the plasmid pSVbeta were grown in three medium formulations to assess their effects on the characteristics of supercoiled plasmid DNA production for plasmid-based gene therapy. A semi-defined medium containing casamino acids (SDCAS) was found to support higher cell densities and higher plasmid stability than a similar medium containing soya amino acids (SDSOY) or Luria-Bertani medium (LB). Differences were observed in the cell harvest characteristics, plasmid DNA primary recovery, plasmid DNA yield and quality between cells grown on LB and on SDCAS medium. Cells grown on SDCAS medium were more difficult to resuspend after harvest than those grown in LB medium and were less susceptible to alkaline lysis. The plasmid DNA content from SDCAS was predominantly supercoiled and was less contaminated by chromosomal DNA than plasmid DNA extracts derived from cells grown on LB medium. It was hypothesised that the different carbon:nitrogen ratio (C:N) of the medium may have been responsible for changing the cell wall polysaccharide composition resulting in the change in cell harvest and lysis characteristics. Results indicated that changing the C:N ratio of SDCAS medium between 1.21:1 and 12.08:1 resulted in no alteration in cell wall polysaccharide composition or in cell susceptibility to chemical lysis or physical breakage. Plasmid DNA yields increased ten-fold with ten-fold increase in the C:N ratio of SDCAS medium.

Rapid quantitation and monitoring of plasmid DNA using an ultrasensitive DNA-binding dye.

A sensitive fluorescence-based method for monitoring plasmid DNA during production was investigated. This simple method of assaying for plasmid DNA allows rapid monitoring of plasmid yields from a recombinant Escherichia coli fed-batch fermentation. The assay has several advantages over traditional methods of plasmid DNA measurement. The fluorescent dye is highly specific and can measure total plasmid DNA concentration in about 5 min. The assay is sensitive over a wide range of plasmid concentrations of between 15 and 280 ng/mL, even in the presence of impurities that occur within alkaline lysate preparations. The technique can also be applied to monitoring fermentation and downstream purification steps.

Factors determining more efficient large-scale release of a periplasmic enzyme from E. coli using lysozyme.

Large scale use of lysozyme for periplasmic release has been impeded by the cost of the pure enzyme and its subsequent presence as a contaminant in later downstream processing steps. In this paper, we discuss the use of lysozyme for pilot scale recovery of a periplasmic enzyme from E. coli. The effects of concentration of sucrose, lysozyme and cells on periplasmic enzyme release were examined. Lysozyme concentration can be reduced 5-fold from previous reports and a reduction in sucrose concentration from 20 to 15% (w/v) allows an improvement in centrifugal harvesting by reducing viscosity. High levels of release were still achieved using this technique and further improvements in yield were obtained by optimising other components of the releasing mixture. Results show that some release is still achieved in circumstances where no lysozyme use is possible. Results also indicate that a substantial proportion (up to 70%) of lysozyme remains bound to the cellular debris after its action and is removed with this material.

Factors affecting the fermentative production of a lysozyme-binding antibody fragment in Escherichia coli.

Fed-batch fermentation for production of a single-chain Fv antibody fragment (scFv) expressed as a recombinant periplastic protein from Escherichia coli was investigated. A high cell density of 50 g dry cell weight per liter was routinely achieved in a 14-L vessel by controlled exponential feeding of glucose to impose a constant specific growth rate. Following biomass accumulation, induction of the tac promoter by addition of IPTG was accompaied by a linear feed of yeast extract. The concentration of yeast extract feed was found to be highly influential upon both concentration and location of active product. Although scFv fragments were specifically targeted to the periplasmic space, at yeast extract feed rates of 0.72 g/h the final location was largely extracellular (68% to 79%). Total concentrations (extracellular + periplasmic) were of the order of 5 to 8 mg/L. A ten-fold increase in yeast extract supply increased total scFv concentration to almost 200 mg/L and 78% of this yield was retained in the periplasm. Control of such leakage of the recombinant product is fundamental to process design of downstream operations for product recovery.

Production of cadmium sulphide microcrystallites in batch cultivation by Schizosaccharomyces pombe.

Cadmium sulphate was added to separate batch cultures of Schizosaccharomyces pombe during different growth phases to determine the effect on cadmium sulphide microcrystallite production. Exit gas analysis was used to determine the impact on metabolism. Addition during the early-exponential growth phase resulted in an immediate intracellular uptake of cadmium, followed by rapid efflux from the cells, permanent reduction in cell metabolism and a lower intracellular inorganic sulphide content. This response was not suitable for cadmium sulphide microcrystallite production. Stationary phase cultures did not induce cadmium sulphide microcrystallite production. However, the addition of cadmium sulphate to a culture during the mid-exponential growth phase increased the intracellular cadmium and inorganic sulphide concentrations for approximately 8 h before reaching a saturation level for the cell. This resulted in a significant level of cadmium sulphide microcrystallite production.

Rheologies and morphologies of three actinomycetes in submerged culture.

The broth rheologies and morphologies of three actinomycetes (Saccharopolyspora erythraea, Actinomadura roseorufa, and Streptomyces rimosus) in submerged culture have been examined. The rheology of all the broths became pseudoplastic as soon as significant growth occurred with the power law index, n, falling to 0.20 to 0.25. The consistency index, K, rose with biomass concentration although in some instances it fell later in the fermentation. The mean main hyphal lengths of all cultures were in the range, 15 to 25 mum, and did not alter greatly even when large changes in K were occurring. (c) 1995 John Wiley & Sons, Inc.

Estimation of cell volume and biomass of penicillium chrysogenum using image analysis.

A methodology for the estimation of biomass for the penicillin fermentation using image analysis is presented. Two regions of hyphae are defined to describe the growth of mycelia during fermentation: (1) the cytoplasmic region, and (2) the degenerated region including large vacuoles. The volume occupied by each of these regions in a fixed volume of sample is estimated from area measurements using image analysis. Areas are converted to volumes by treating the hyphae as solid cylinders with the hyphal diameter as the cylinder diameter. The volumes of the cytoplasmic and degenerated regions are converted into dry weight estimations using hyphal density values available from the literature. The image analysis technique is able to estimate biomass even in the presence of nondissolved solids of a concentration of up to 30 gL(-1). It is shown to estimate successfully concentrations of mycelia from 0.03 to 38 gL(-1). Although the technique has been developed for the penicillin fermentation, it should be applicable to other (nonpellected) fungal fermentations.