Modeling of scale-down effects on the hydrodynamics of expanded bed adsorption columns

Expanded-bed adsorption (EBA) is a technique for primary recovery of proteins starting from unclarified broths. This process combines centrifugation, concentration, filtration, and initial capturing of the proteins in a single step. An expanded bed (EB) is comparable to a packed bed in terms of separation performance but its hydrodynamics are that of a fluidized bed. Downstream process development involving EBA is normally carried out in small columns to minimize time and costs. Our purpose here is to characterize the hydrodynamics of expanded beds of different diameters, to develop scaling parameters that can be reliably used to predict separation efficiency of larger EBA columns. A hydrodynamic model has been developed which takes into account the radial liquid velocity profile in the column. The scale-down effect can be characterized in terms of apparent axial dispersion, D(axl,app), and plate number, N(EB), adapted for expanded bed. The model is in good agreement with experimental results obtained from 1- and 5-cm column diameters with buffer solutions of different viscosities. The model and the experiments show an increase of apparent axial dispersion with an increase in column diameter. Furthermore, the apparent axial dispersion is affected by an increase in liquid velocity and viscosity. Supported by visual observations and predictions from the model, it was concluded that operating conditions (liquid viscosity and superficial velocity) resulting in a bed-void fraction between 0.7 and 0.75 would provide the optimal separation efficiency in terms of N(EB).     

Evaluation of the effect of in-bed sampling on expanded bed adsorption

An expanded bed adsorption (EBA) column (5 cm diameter) has been modified to allow the abstraction of liquid samples from various positions along the height of an expanded bed. As the adsorbent particles were fluidized, in-bed monitoring of key component concentrations during feedstock application, washing and elution was achieved by the withdrawal of liquid samples from the voids within the expanded bed through ports along the wall of the column. Component levels in the withdrawn streams can be assayed using on-line analytical chromatography or samples can be collected and assayed off-line. On-line monitoring can be used to control the duration of the loading stage and as a tool to provide information about the hydrodynamic and adsorption/desorption processes that occur during expanded bed adsorption. Studies of residence time distributions indicated that the modifications to the column do not significantly affect liquid dispersion. Using the adsorption of glucose-6-phosphate dehydrogenase from yeast homogenate on Streamline DEAE as a model system, comparison of breakthrough curves for runs when in-bed monitoring was and was not performed also suggested that separation efficiency is not appreciably affected by in-bed sampling.     

The effect of column verticality on separation efficiency in expanded bed adsorption

The effect of column verticality on liquid dispersion and separation efficiency in expanded bed adsorption columns was investigated using 1 and 5 cm diameter columns. Column misalignment of only 0.15° resulted in the reduction of the Bodenstein number from 140 to 50 for the 1 cm dia. column and from 75 to 45 for the 5 cm dia. column. This degree of misalignment was not detectable by visual assessment of adsorbent particle movement within the column. Depending on the relative importance of transport limitations, kinetic limitations and dispersion to any specific separation, this increase in dispersion with column alignment can significantly affect separation efficiency. Pure protein breakthrough profiles resulting from the application of bovine serum albumin onto STREAMLINE Q XL demonstrated that, at 10% breakthrough, 7.8% more protein could be applied to a vertical 1 cm dia. column compared to the same column misaligned by 0.15°. When an unclarified yeast homogenate was applied to a 1 cm dia. vertical column packed with STREAMLINE DEAE, 10% breakthrough of glucose-6-phosphate dehydrogenase (G6PDH) corresponded to a load 55% greater compared to the same column aligned 0.185° off-vertical. The G6PDH breakthrough curves for vertical and 0.15° off-vertical runs performed using a 5 cm column were essentially indistinguishable.     

Determining particle density distribution of expanded bed adsorbents

This study presents an experimental approach to measure the density distribution of expanded bed adsorption (EBA) matrices. We report on the use of a series of solutions of caesium trifluoroacetate (CsTFA) of varying density spun in a laboratory centrifuge so as to separate representative matrix samples on the basis of bead density. Mass data was used to plot a decumulative density distribution for the matrix. By performing laser light scattering-based measurements on the same samples of matrix the variation in particle size with density was determined. Particle settling velocity distributions were then calculated using these data and compared with a settling velocity distribution calculated on the basis of an assumed constant bead density. The study demonstrates a reliable and simple method for the characterisation of matrix density distribution. For the case of the Streamline matrices tested the particle size distribution is constant with varying bead density. Bead densities varied from 1.5 to 2.1 g/cm3 in the CsTFA solutions. These were then adjusted using bead porosity to give a density range of 1.11-1.33 g/cm3 in aqueous buffer (assumed 1.0 g/cm3) The differences in resultant settling velocity distributions when based upon measured density distribution than when based upon an assumed mean density value were shown to be insignificant. This result confirms experimentally that an assumption of a single constant mean density for EBA particles is acceptable for hydrodynamic modelling and performance prediction purposes.     

Interfacing Pichia pastoris cultivation with expanded bed adsorption

For improved interfacing of the Pichia pastoris fed-batch cultivation process with expanded bed adsorption (EBA) technique, a modified cultivation technique was developed. The modification included the reduction of the medium salt concentration, which was then kept constant by regulating the medium conductivity at low value (about 8 mS/cm) by salt feeding. Before loading, the low conductivity culture broth was diluted only to reduce viscosity, caused by high cell density. The concept was applied to a one-step recovery and purification procedure for a fusion protein composed of a cellulose-binding module (CBM) from Neocallimastix patriciarum cellulase 6A fused to lipase B from Candida antarctica (CALB). The modified cultivation technique resulted in lower cell death and consequently lower concentration of proteases and other contaminating proteins in the culture broth. Flow cytometry analysis showed 1% dead (propidium-stained) cells compared to 3.5% in the reference process. During the whole process of cultivation and recovery, no proteolysis was detected and in the end of the cultivation, the product constituted 87% of the total supernatant protein. The lipase activity in the culture supernatant increased at an almost constant rate up to a value corresponding to 2.2 g/L of CBM-CALB. In the EBA process, no cell-adsorbent interaction was detected but the cell density had to be reduced by a two-times dilution to keep a proper bed expansion. At flow velocity of 400 cm/h, the breakthrough capacity was 12.4 g/L, the product yield 98%, the concentration factor 3.6 times, the purity about 90%, and the productivity 2.1 g/L x h.     

Extreme scale-down of expanded bed adsorption: Purification of an antibody fragment directly from recombinant E. coli culture

Scale-down is a methodology that combines the use of very small volumes of process fluid in dedicated devices to predict accurately the behaviour of process-scale biotechnological unit operations and for the production of comparable material for use in further devices which, taken together, facilitate the mimic of a complete full-scale process. This article provides the rationale behind the development of a small-scale mimic and demonstrates the use of a highly scaled-down expanded bed to predict hydrodynamic, kinetic, and adsorptive performance using less than 5-mL sample volumes. Data acquired on a specially developed 1.9 mm ID column was compared with that obtained in a standard 25 mm ID column. A homogenised E. coli system expressing an antibody fragment (F(ab)) adsorbed onto an rProtein A matrix was used to characterise the full adsorptive performance. Breakthrough curve studies using BSA in buffer were used to characterise binding kinetics. Performance at the two scales was comparable both in terms of expansion, axial dispersion, binding isotherms, and elution behaviour of the antibody fragment. The eluted F(ab) material was further purified by ion exchange chromatography to demonstrate the similarity between the profile of the product material obtained at both scales. The high level of scale-down (approximately 200-fold) provides for rapid process evaluation early in development, where material is at a premium and where a fast appreciation of the likely merits of one process strategy will lead to greater confidence in process selection and more robust flowsheets.     

Single-step recovery of ephedrine hydrochloride from raw materials using expanded bed adsorption

Expanded bed adsorption, using a cation resin 001 x 7 Styrene-DVB, was used to recover and purify ephedrine hydrochloride from a powdered herb. The axial liquid-phase dispersion coefficient was about 10(-5) m(2) s(-1) and the recovery yield and purification reached 86% and 22, respectively. Compared with using conventional extraction with dimethylbenzene, this method is safer and also more efficient.     

Expanded bed adsorption: elution in expanded bed mode

Elution in expanded bed mode has been investigated in the expanded bed adsorption process. Elution was performed at different sample loads and at different liquid velocities using bovine serum albumin as a model. The effect on mixing in the liquid phase and on the volume of the eluted peak were determined. Mixing in the liquid phase was almost unaffected when elution was performed at 100 cm/h, regardless of sample load. However, mixing increased significantly when elution was carried out at high liquid velocities (300 cm/h) at high sample loads. The eluted peak volume increased with liquid velocity and increased sample load. It was approx. 80% higher in expanded bed mode than in packed bed from an adsorbent completely saturated with protein eluted at 300 cm/h.     

Buoyancy-induced mixing during wash and elution steps in expanded bed adsorption

Buoyancy-induced mixing occurs during expanded bed adsorption processes when the feed stream entering the bottom of the system has a lower density than that of the fluid above it. In the absence of a headspace, mixing in the expanded bed can be modeled as a single, well-mixed vessel, with first-order dynamics. In the presence of a headspace, the system exhibits second-order dynamics for the densities typically encountered in protein chromatography, and can be modeled as two well-mixed vessels (the expanded bed and the headspace) arranged in series. In this paper, the mixing dynamics of the expanded bed are described and a mathematical model of the system is presented. Experimental measurements of density changes during the dilution of sucrose and salt solutions in a STREAMLINE 25 column are presented. These show excellent agreement with predictions using the model. A number of strategies for wash and elution in expanded mode, both in the presence and absence of headspace, are discussed.     

Pilot scale recovery of monoclonal antibodies by expanded bed ion exchange adsorption

The aim of the investigations was to estimate the scale up properties of an efficient chromatographic first capture step for the recovery of murine IgG₁ from undiluted and unclarified hybridoma cell culture broth using an ion exchange matrix in expanded bed mode. The tested new sulfopropyl-based ion exchange matrix (Streamline^TM SP XL, Amersham Pharmacia Biotech) stands out due to its enhanced capacity compared to its precursor (Streamline^TM SP). Defining the working pH in preliminary electrophoretic analyses (titration curve, SDS-PAGE) and small-scaled chromatographic binding studies showed, that the optimal value for the IgG purification was pH 4.6, where a co-chromatography of the medium supplement albumin (500 mg l^-1, pI = 4.8) could not be avoided. Further scouting experiments dealt with the dynamic capacity of the matrix, which was evaluated by frontal adsorption analysis. In packed bed mode no break-through of the target protein was achieved even after 6.5 mg IgG per ml matrix were applied. These results could not be reproduced in expanded bed mode with cell-free supernatant, where the dynamic capacity was found to be only 1.5 mg IgG/ml SP XL. Processing cell-containing broth resulted in an additional decrease of the value down to 0.5 mg ml^-1, presumably caused by the remarkable biomass adsorption to the matrix. The search for the reasons led to the examination of the hydrodynamic conditions. Buffer experiments with a tracer substance (acetone) pointed out, that the flow in expanded bed was significantly more influenced by back-mixing effects and channel formations than in packed bed. These effects could be compensated with an enhanced viscosity of the liquid phase, which was achieved by the addition of glucose. As a result of the improved hydrodynamic conditions in the expanded bed, the dynamic capacity could be increased from 0.5 to more than 4.5 mg IgG/ml matrix for the processing of cell culture broth with 400 mM glucose. Finally, the scale up from a Streamline^TM 25 to a Streamline^TM 200 column was performed under conditions, which proved to be optimal: 100 L of unclarified hybridoma broth were concentrated with a binding rate of 95% in less than 3.5 hours. Loading the column no break-through of the target protein was achieved. However, the eluate still contained debris and cells, which points out the major disadvantage of the method: the biomass attachment to the matrix.     

Anion exchange purification of plasmid DNA using expanded bed adsorption

Recent developments in gene therapy with non-viral vectors and DNA vaccination have increased the demand for large amounts of pharmaceutical-grade plasmid DNA. The high viscosity of process streams is of major concern in the purification of plasmids, since it can cause high back pressures in column operations, thus limiting the throughput. In order to avoid these high back pressures, expanded bed anion exchange chromatography was evaluated as an alternative to fixed bed chromatography. A Streamline 25 column filled with 100 ml of Streamline QXL media, was equilibrated with 0.5 M NaCl in TE (10 mM Tris, 1 mM EDTA, pH=8.0) buffer at an upward flow of 300 cmh^-1, E. coli lysates (obtained from up to 3 liters of fermentation broth) were injected in the column. After washing out the unbound material, the media was allowed to sediment and the plasmid was eluted with 1 M NaCl in TE buffer at a downward flow of 120 cmh^-1. Purification factors of 36±1 fold, 26±0.4 plasmid purity, and close to 100% yields were obtained when less than one settled column volume of plasmid feed was injected. However, both recovery yield and purity abruptly decreased when larger amounts were processed–values of 35±2 and 5±0.7 were obtained for the recovery yield and purity, respectively, when 250 ml of feedstock were processed. In these cases, gel clogging and expansion collapse were observed. The processing of larger volumes, thus larger plasmid quantities, was only possible by performing an isopropanol precipitation step prior to the chromatographic step. This step led to an enhancement of the purification step.     

Factors affecting dispersion in expanded bed chromatography

Factors affecting the dispersion of solutes in expanded bed chromatography were experimentally investigated to characterize the behavior in small columns. Pulse response curves were measured with a vitamin B¹² tracer, and HETP (height equivalent to a theoretical plate) values were calculated from peak variance and retention time. Approximately 15 min were required to attain a stable steady state expanded bed height with constant HETP values. HETP values ranged from 0.8 to 1.6 cm and did not change appreciably with the degree of expansion (1.5–3.5 fold), column diameter (1.6 and 2.6 cm), column temperature (293–308 K) or settled bed height (ca. 4–11 cm). A very small column (1.6 cm diam. and 4.2 cm-settled bed height) was successfully expanded and axial mixing measured could be useful for conducting scale down experiments.     

Design of expanded bed supports for the recovery of plasmid DNA by anion exchange adsorption

In this study we detail the rational design of new chromatographic adsorbents tailored for the capture of plasmid DNA. Features present on current chromatographic supports that can significantly enhance plasmid binding capacity have been identified in packed bed chromatography experiments and blueprints for improved expanded bed adsorbents have been put forward. The characterisation and testing of small (20-40 m) high density (>3.7 g cm^–3) pellicular expanded bed materials functionalised with various anion exchange structures is presented. In studies with calf thymus DNA, dynamic binding capacities of 1.2 and 3.4 mg ml^–1 were recorded for prototype diethylaminoethyl-and polyethylene imine-linked adsorbents which were respectively 25 and 70 fold higher than those of equivalently derivatised commercial expanded bed materials. The prototype polyethylene imine-coupled material exhibited severe sensitivity to inter-particle bridging by nucleic acid polymers, gave low DNA recoveries (<37%) and proved difficult to regenerate. In contrast, few operational difficulties were experienced with the diethylaminoethyl-linked prototype adsorbent and successful high capacity (>0.8 mg ml^–1) capture of plasmid DNA from crude neutralised E. coli lysate was demonstrated.     

Partial purification of nattokinase from Bacillus subtilis by expanded bed adsorption

Nattokinase was purified from unclarified Bacillus subtilisculture filtrate using an expanded bed with a purification factor of 8.2 and at a yield of 95%. The optimal pHs for adsorption and elution were 6.0 and 7.0, respectively. The expanded bed route shortened the process time by 8–10 h and increased the yield by 50% when compared with the conventional method.     

Routine manufacture of recombinant proteins using expanded bed adsorption chromatography

Two different recombinant human proteins were purified directly from Pichia pastoris whole cell fermentation broth, containing 30–44% biomass (wet weight percent), by strong cation exchange expanded bed adsorption chromatography. Expanded bed adsorption chromatography provided clarification, product purification and product concentration in a single unit operation at large scale (2000-l nominal fermentation volume). The efficiency of expanded bed adsorption chromatography resulted in a short process time, high process yield, and limited proteolytic degradation of the target proteins. The separations were operated using a 60-cm (d) column run at 14 l/min. For one protein, expanded bed adsorption chromatography resulted in an average product recovery of 113% (relative to fermentation supernatant) and a purity of 89% (n=10). For the other protein, the average product recovery was 99% (relative to fermentation supernatant) and the purity was 62.1 (n=10). Laboratory experiments showed that biomass reduced product dynamic binding capacity for protein 2.     

Physical and biochemical characterization of a simple intermediate between fluidized and expanded bed contactors

Physical and biochemical comparison has been made of the performance of a simple fluidised bed contactor and a commercial expanded bed contactor, characterised by identical dimensions, and operated at various settled bed heights with two anion exchange adsorbents. The contactors were tested with various feedstocks comprising bovine albumin in the absence and presence of 20 g dry cell weight biomass litre^-1. Earlier classification of the simple contactor as a single-stage, well mixed fluidised bed is reviewed. The relative merits of STREAMLINE DEAE and DEAE Spherodex LS as fluidisable, anion exchange adsorbents are discussed.     

On-column refolding of recombinant human interferon-gamma inclusion bodies by expanded bed adsorption chromatography

A refolding strategy was described for on-column refolding of recombinant human interferon-gamma (rhIFN-gamma) inclusion bodies by expanded bed adsorption (EBA) chromatography. After the denatured rhIFN-gamma protein bound onto the cation exchanger of STREAMLINE SP, the refolding process was performed in expanded bed by gradually decreasing the concentration of urea in the buffer and the refolded rhIFN-gamma protein was recovered by the elution in packed bed mode. It was demonstrated that the denatured rhIFN-gamma protein could be efficiently refolded by this method with high yield. Under appropriate experimental conditions, the protein yield and specific activity of rhIFN-gamma was up to 52.7% and 8.18 x 10(6) IU/mg, respectively.     

Cell/adsorbent interactions in expanded bed adsorption of proteins

Expanded bed adsorption (EBA) is an integrated technology for the primary recovery of proteins from unclarified feedstock. A method is presented which allows a qualitative and quantitative understanding of the main mechanisms governing the interaction of biomass with fluidised resins. A pulse response technique was used to determine the adsorption of various cell types (yeast, Gram positive and Gram negative bacteria, mammalian cells and yeast homogenate) to a range of commercially available matrices for EBA. Cells and cell debris were found to interact with the ligands of agarose based resins mainly by electrostatic forces. From the adsorbents investigated the anion exchange matrix showed the most severe interactions, while cation exchange and affinity adsorbents appeared to be less affected. Within the range of biologic systems under study E. coli cells had the lowest tendency of binding to all matrices while hybridoma cells attached to all the adsorbents except the protein A affinity matrix. The method presented may be employed for screening of suitable biomass/adsorbent combinations, which yield a robust and reliable initial capture step by expanded bed adsorption from unclarified feedstock.     

The influence of complex biological feedstock on the fluidization and bed stability in expanded bed adsorption

The stability of expanded bed adsorption systems (EBA) was studied in biomass containing culture broth by residence time distribution (RTD) experiments, using pulse inputs of fluorescent molecules as tracers. Different commercial adsorbents (Streamline DEAE, SP, Phenyl, Chelating, and AC) were tested at various biomass concentrations (2.5–12 %, wet weight) of whole (Saccharomyces cerevisiae) yeast, yeast cell homogenate, and Escherichia coli homogenate. Analyzing the RTD according to the PDE model (PDE: axially dispersed plug‐flow exchanging mass with stagnant zones) allowed the calculation of three parameters: the number of transfer units for mass exchange between mobile and stagnant fraction (N), the Peclet number for overall axial dispersion (P), and the mobile fraction of the liquid in axially dispersed plug flow (φ). When fluidization was performed in particle‐free buffer the normalized response signal (after perfect input pulse) was symmetric (N:0; P: 50–100; φ: 1), thus, demonstrating the formation of a homogeneous fluidized (expanded) bed. Upon application of suspended biomass the RTD was skewed, depending on the adsorbent used and the type and level of biomass present in the sample. This situation leads to three different characteristic pictures: the well‐fluidized system (N: ≥ 7–10; P: ≥ 40; φ: 0.80–0.90), the system exhibiting bottom channeling (N: < 1–2; P: ≥ 40; φ: 0.5–0.7) and, the system where extensive agglomeration develops (N: 4–7; P: 20–40; φ: < 0.5). These results demonstrate that changes in the hydrodynamics of EBA already take place in the presence of moderate concentrations of biomass. Furthermore, those changes can be quantitatively described mainly in terms of the fraction of stagnant zones in the system, which are formed due to the interaction of biomass and adsorbent. The technique described here can be used to evaluate a certain combination of adsorbent and biomass with regard to its suitability for expanded bed adsorption from whole broth. © 1999 John Wiley & Sons, Inc. Biotechol Bioeng 64: 484–496, 1999.     

Removal of poly-histidine fusion tags from recombinant proteins purified by expanded bed adsorption

Enzymatic methods have been used to cleave the C- or N-terminus polyhistidine tags from histidine tagged proteins following expanded bed purification using immobilized metal affinity chromatography (IMAC). This study assesses the use of Factor Xa and a genetically engineered exopeptidase dipeptidyl aminopeptidase-1 (DAPase-1) for the removal of C-terminus and N-terminus polyhistidine tags, respectively. Model proteins consisting of maltose binding protein (MBP) having a C- or N-terminal polyhistidine tag were used. Digestion of the hexahistidine tag of MBP-His(6) by Factor Xa and HT15-MBP by DAPase-1 was successful. The time taken to complete the conversion of MBP-His(6) to MBP was 16 h, as judged by SDS-PAGE and Western blots against anti-His antibody. When the detagged protein was purified using subtractive IMAC, the yield was moderate at 71% although the overall recovery was high at 95%. Likewise, a yield of 79% and a recovery of 97% was obtained when digestion was performed with using "on-column" tag digestion. On-column tag digestion involves cleavage of histidine tag from polyhistidine tagged proteins that are still bound to the IMAC column. Digestion of an N-terminal polyhistidine tag from HT15-MBP (1 mg/mL) by the DAPase-I system was superior to the results obtained with Factor Xa with a higher yield and recovery of 99% and 95%, respectively. The digestion by DAPase-I system was faster and was complete at 5 h as opposed to 16 h for Factor Xa. The detagged MBP proteins were isolated from the digestion mixtures using a simple subtractive IMAC column procedure with the detagged protein appearing in the flowthrough and washing fractions while residual dipeptides and DAPase-I (which was engineered to exhibit a poly-His tail) were adsorbed to the column. FPLC analysis using a MonoS cation exchanger was performed to understand and monitor the progress and time course of DAPase-I digestion of HT15-MBP to MBP. Optimization of process variables such as temperature, protein concentration, and enzyme activity was developed for the DAPase-I digesting system on HT15-MBP to MBP. In short, this study proved that the use of either Factor Xa or DAPase-I for the digestion of polyhistidine tags is simple and efficient and can be carried out under mild reaction conditions.