- Email: [email protected]
Expanded-bed adsorption process for protein capture
EXPANDED-BED ADSORPTION PROCESS FOR PROTEIN CAPTURE JOSEPH SHILOACH Biotechnology Unit, National Institute of Diabetes and Digestive and Kidney Disease, National Institutes of Health, Bethesda, Maryland 20892
ROBERT M. K E N N E D Y Separations Group, Amersham Pharmacia Biotech, Piscataway, New Jersey 088SS
I. II. III. IV. V. VI.
INTRODUCTION PRINCIPLES OF EXPANDED-BED OPERATION EXPERIMENTAL STRATEGY INSTRUMENTATION MATRICES APPUCATIONS A. Capturing Extracellular Proteins B. Capture of Intracellular Proteins VII. DISCUSSION AND CONCLUSIONS REFERENCES
INTRODUCTION
The complete process for obtaining pure proteins can be divided into three main steps: capture, purification, and poHshing. The first step is the immobihzation of the target protein onto some adsorptive surface, and it can be viewed as a combination of clarification, concentration, stabilization and initial purification. Because the starting protein solution (feedstock) is usually crude, it is essential to clarify the solution. The traditional or conventional approach involves centrifugation, microfiltration, ultrafiltration, or diafiltration before the target protein solution can be loaded on an adsorbing material, utilizing packed-bed chromatography. The clarification step is a demanding operation, and is particularly difficult v^hen processing large quantities of microorganisms, especially disrupted microorganisms. Highspeed, large-scale centrifugation and microfiltration are the most common processes used to obtain protein solutions that are suitable for packed-bed chromatography; therefore, it is obvious that an approach that eliminates the Separation Science and Technology, Volume 2 Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved.
43 I
432
JOSEPH SHILOACH AND ROBERT M. KENNEDY
clarification step can significantly simplify and improve the overall purification process. Direct adsorption of the protein not only eliminates the clarification step, but also produces a concentrated and partially purified product ready for the next purification step (Fig. 1). Several protein capture procedures, such as batch adsorption, solvent extraction, and expanded-bed adsorption, are protein capture procedures that do not require centrifugation and filtration. In this chapter wt describe the expanded-bed adsorption approach for capturing target proteins. In the expanded-bed mode, the starting protein solution is pumped upw^ard through a bed of adsorbent beads that are constrained by a flow^ adaptor.^ As a result of the upward flow and the properties of the beads, the bed expands as spaces open up between the beads. If the physical properties of the beads are significantly different from those of the particles in the feedstock, the particles can pass freely through the bed without being trapped.
Extracellular
Fermentation
Intracellular
Cell separation ppt (Cells)
Sup.
Clarification
Suspension
Concentration
Cell disruption
Dialysis
Clarification
Column chromatography
Concentration
f
' Dialysis
Column chromatography
.--
F I G U R E I General purification schemes of extracellular and intracellular proteins using traditional and expanded-bed approaches. The solid lines indicate the traditional approach and the broken lines the expanded-bed approach. Using the expanded bed, when the product is extracellular, the fermentation broth Is pumped directly on the column; when the product is intracellular, the suspension of disrupted cells is pumped directly on the column.
433
EXPANDED-BED ADSORPTION
An effective process, which means the formation of a stable expanded bed, depends on parameters such as the viscosity, ionic strength, sohds content, and pH of the feedstock as well as the linear flow rate. Recently, the expanded-bed approach has been successfully used to combine clarification, concentration, and initial purification. The references cited in two recent reviews provide a good summary of these applications.^'^
II. PRINCIPLES OF EXPANDED-BED OPERATION The expanded bed is, in principle, similar to a fluidized bed, a common unit-operation in the chemical industry.^ However, in the expanded-bed method, mixing of the adsorbent material is minimal during the operation, whereas a fluidized bed is turbulent. This means that an expanded bed is more effective for adsorption and elution processes than the fluidized bed. A complete review of all engineering aspects, comparing expanded beds and fluidized beds, can be found in Thommes."^ The expanded-bed adsorption operation is illustrated in Fig. 2. The overall operation is comprised of several phases. In the first phase, the adsorbent material is expanded and equiUbrated by applying an upward liquid flow to the column. To allow for sufficient contact time and efficient binding of the target molecule, the expansion should be three times that of the sedimented bed ("expansion ratio"), to a height of approximately 50 cm. A stable bed is formed when the adsorbent particles achieve equilibrium between particle sedimentation velocity and upward liquid flow velocity. In the second phase, the sample is applied to the expanded adsorbent. The crude, unclarified protein solution of intact or disrupted biomass is pumped upward on the column. In a well-defined process, the expanded adsorbent will remain stable and will not change its expansion ratio. However, if the
Before start-up sedimented adsorbent FIGURE 2
Expansion and equilibration of the adsorbent
Application of feed, followed by washing
Schematic presentation of the steps of expanded-bed adsorption.
Elution in packed bed
434
JOSEPH SHILOACH AND ROBERT M. KENNEDY
process is not defined, there is a chance that the expanded adsorbent will not remain stable and will collapse or expand to the top of the column. More details are provided later in Section VI, "Application." During loading, the target proteins are bound to the adsorbent matrix, while cell debris and contaminants pass through the bed and the upper flow adaptor. The third phase is the washing of the expanded adsorbent (in an upward flow) using a buffer, most often the loading buffer. During this step, weakly bound materials, such as residual cells, cell debris, and other types of particles, are washed out. Following this step, the flow through the column is stopped and the adsorbent material settles. The upper flow adaptor is lowered to the surface of the settled adsorbent and the flow is reversed, going downward. The captured proteins are eluted from the sedimented bed, using appropriate elution solutions. The final phase, as in other adsorption processes, is the washing and regeneration of the adsorbing material. Detailed information is provided by the manufacturer.^ It is possible to use the adsorbent for a large number of runs without losing its functionality. As indicated, expanded-bed adsorption is based on controlled, stable fluidization, thus combining the hydrodynamic properties of a fluidized bed with the chromatographic properties of a packed bed. Stable fluidization with minimal backmixing results in a performance that is similar to that of a traditional plug flow column. The stability of the bed arises from the special design of the distributor plate at the base of the column and from two properties of the adsorbent: particle size distribution and particle density distribution. The distributor plate generates the required pressure drop and directs the flow only in a vertical direction, thus eliminating any radial flow that might create turbulence through the column. The size distribution and the density distribution of the beads ensure minimal local mixing. Table 1 summarizes the work of Hanson^ and illustrates the distribution of particle size and density in the bed. Particle distribution is described from the top of the column down: the top 30 mL contain 12% of the particles with an average size of 144 ^tm and an average density of 1.15 gm/mL, while the bottom 104 mL, 4 1 % of the expanded bed volume, contain particles with an average size of 238 fim and an average density of 1.19 g/mL. The polydispersity of the beads positions them at specific heights in the expanded bed; the smaller, lighter particles move to positions at the top and the larger.
TABLE I
Distribution of Beads in an Expanded Bed°
Sedimented gel volume (mL)
%
Average particle size ()Jim)
Density (gm / mL)
30 49 72 104
12 19 28 41
144 164 186 238
1.15 1.16 1.17 1.19
Degree of expansion: 2.5 folds in water, at 300 cm/hr, in a 5.0 cm diameter column.
435
EXPANDED-BED ADSORPTION
o o
o
o o o
Upper adapter o o o o O
no o
o , '^o
Qp^ O O ^
Lower adapter
booOocn I Flow
FIGURE 3
Packed-bed (a) and expanded-bed (b) columns.
heavier particles move to the bottom, resulting in a stable expansion (Fig. 3). In other v^ords, the beads find their ideal position in the column, v^hich is the reason for the lowr axial dispersion in the expanded-bed operation. The high density of the beads is necessary in order to be able to run the expanded bed at high flow velocities, thus achieving higher productivity.
III. EXPERIMENTAL STRATEGY As with other purification procedures, to develop an expanded-bed procedure, one must first follow an experimental strategy composed of method scouting, method optimization, process verification, and production. The purpose of method scouting is to define the most suitable adsorbent and the optimal conditions for binding and elution. The adsorbent of choice should be the one showing the strongest binding of the target protein, while hardly binding contaminating proteins, demonstrating the highest selectivity a n d / o r the highest capacity for the protein of interest. These initial determinations are done using packed-bed chromatography with clarified feedstock. The conditions must be verified and adjusted using the expanded bed with unclarified feedstock. The feedstock flows through the expanded bed at the relatively high flow rate of 300 to 400 cm/hr; therefore, it is important to ensure that the binding parameters of the expanded bed with the unclarified feedstock are similar to those determined using the packed bed process. Figure 4 shows breakthrough capacity curves of three different columns, one packed-bed and two expanded-bed that represent laboratory and pilot scale productions. Increasing amounts of BSA were loaded into the columns and the fraction of the unbound BSA was monitored. Please note the similar
436
JOSEPH SHILOACH AND ROBERT M. KENNEDY
C/Co Packed XK16
1.0
Streamline® 50 Streamline 200
0.8
0.6
h
0.4
0.2
0.0 00
20
40
60
Applied BSA (mg/ml adsorbent) FIGURE 4 Breakthrough curves for BSA on STREAMLINE DEAE. The breakthrough capacity is determined using frontal analysis.
shapes of the breakthrough curves for the packed-bed and expanded-bed modes, as well as for the different sizes of expanded-bed columns with internal diameters of 50 and 200 mm. This similarity demonstrates that the adsorption performance is the same in expanded-bed and packed-bed modes. The special expanded-bed column (STREAMLINE, Amersham Pharmacia Biotech, Sweden) can be scaled up while maintaining the same performance. Once a suitable maximum loading capacity is defined, it is suggested that a safety margin of 75% of the breakthrough capacity be applied, to compensate for sample variability that can affect the binding capacity. The translation of the packed-bed conditions to expanded-bed operation is done with crude, unclarified feedstock and an expanded-bed column with a diameter of 2.5 or 5 cm. Since proper operation of the expanded bed depends on stable expanded adsorbent, and the stability of the adsorbent depends on the viscosity, the ionic strength and the flow velocity of the loading solution, these parameters are adjusted.
437
EXPANDED-BED ADSORPTION
Cleaning-in-place (CIP) procedures^ are important for expanded-bed operations due to the high debris load. Most other chromatographic matrices are never exposed to such levels of contamination. It is also important to verify that bed characteristics have not been altered after the CIP. After establishing process parameters, pilot-scale work may begin with a pilot-scale column (e.g., 20 cm in diameter). For some applications, this column may be suitable for the final production, since such a column has a capacity for 4.7 liters of matrix. Further scale-up is accomplished by increasing the column diameter while maintaining the sedimented bed height, the flow velocity, and the expanded bed height. Larger columns currently in use have a diameter of 1 m.
IV. INSTRUMENTATION The basic configuration and optional setup of the expanded-bed operation is shown in Fig. 5. In addition to the special column, there is a need for two pumps and a set of valves to control the flow to and from the column. Columns for expanded-bed adsorption are commercially available from vari-
Hydraulic liquid
Waste F I G U R E 5 Basic configuration of expanded-bed column operation. Pump 2 pushes down the flow adaptor, pump I controls the upward flow of the sample and the various buffers. Valve VI is a four-way valve directing the flow to and from the column.
438
JOSEPH SHILOACH AND ROBERT M. KENNEDY
ous sources. The two main features that characterize the STREAMLINE columns are the unique design of the liquid distribution system and a moveable adaptor. Columns range in size, from 2.5 cm in diameter used for method development, 5 and 10 cm used for process development, and up to 1200 cm used for production purposes. Columns ranging from 2.5 to 200 cm are built from glass, while the larger ones are made of stainless steel. Another company, Upfront (Denmark), offers an expanded-bed column with a special mixing device at its base. As mentioned, the expanded-bed operation requires two pumps: one for pumping the feedstock, the wash, and the elution solutions on the column, and the other for lowering the flow adaptor. To control flow direction, four-way valves and two-way valves are needed. Manual and automated control systems are available for operating an expanded-bed column.
V. MATRICES
Gel matrices for expanded-bed adsorption are commercially available, and their fluidization properties are reviewed by Thommes. 4 The adsorbent (e.g., STREAMLINE DEAE) exhibits a Gaussian-like distribution of particle sizes and particle densities. As explained previously, this particle polydispersity is an important factor that contributes to the stability of the expanded bed. The mean particle size of STREAMLINE beads is 200 ~m compared to approximately 90 ~m for matrices used in packed-bed chromatography, and the density is about 1.2 g/mL. The various matrices available under the STREAMLINE name are summarized in Tables 2 and 3. The ion-exchangers are based on highly crosslinked 6% agarose modified by including an inert quartz core to give the desired density. In the XL versions of the STREAMLINE matrices, long molecules of dextran are coupled to agarose, and the ion-exchange groups are then attached to these dextran chains. The STREAMLINE protein A is based on highly crosslinked 4% agarose modified by including an inert metal alloy core. Not mentioned in the tables is STREAMLINE phenyl used for hydrophobic interaction chromatography.
VI. APPLICATIONS
Sections I-V addressed the principles of expanded-bed operation, the instrumentation, and the special matrices. This section focuses on applications of the expanded bed, describing actual processes and providing operational guidelines. The expanded-bed adsorption process replaces the centrifugation, clarification, and concentration steps and allows the direct adsorption of the
439
EXPANDED-BED ADSORPTION
T A B L E 2 S T R E A M L I N E Matrices Available for Ion-Exchange Expanded-Bed Adsorption Matrix
SP
SPXL
DEAE
QXL
Type of ion-exchanger Ionic capacity (mmol/mL) Particle size ({xm) Mean particle size (/xm) Approx. mean density (gm/mL) Expansion at 300 c m / h r pH stability^ Long term Short term Working velocity (cm/hr) Capacity (mg/mL) Lysozyme BSA
Strong cation 0.17-0.24 100-300 200 1.2 2-3
Strong cation 0.18-0.24 100-300 200 1.2 2-3
Weak anion 0.13-0.21 100-300 200 1.2 2-3
Strong anion 0.23-0.33 100-300 200 1.2 2-3
4-13 3-14 300
4-13 3-14 300-500
2-13 2-14 300
2-12 2-14 300-500
>60 n.d.
>140 n.d.
n.d. >40
n.d. >110
Total ionic capacity by titration. Long term refers to the pH interval v^here the gel is stable over a long period of time w^ithout adverse effects on its chromatographic performance. Short term refers to the pH interval suitable for regeneration and cleaning. For the SP and the DEAE matrices, breakthrough capacity v^as determined in a STREAMLINE 50 column at a How velocity of 300 c m / h r using a 2.0 m g / m L protein solution in 50 m M sodium phosphate buffer pH 7.5 (lysozyme) and 50 m M Tris-HCl buffer pH 7.5 (BSA). Sedimented bed height w^as 15 cm. For the SPXL and QXL matrices, breakthrough capacity v^as determined in a packed bed, 4.4 mL, at a flov^ velocity of 300 c m / h r using a 2.0 m g / m L solution of protein in 50 m M glycine-HCl buffer pH 9.0 (lysozyme) and 50 m M Tris-HCl buffer pH 7.5 (BSA). Bed height was 10 cm.
protein on a proper matrix. The conditions of a specific process depend on the source and the properties of a particular protein and on its location in the microorganism. The protein can be obtained from prokaryotic and eukaryotic microorganisms (native or recombinant), from mammalian cells and tissues, insect cells, plant tissues, and other sources. In some cases, the protein can be intracellular and in others extracellular. This application section is, therefore, divided into two parts: one covering the capture of extracellular proteins and the other covering the capture of intracellular proteins. In each case, general considerations are given, and several examples of different matrices and different protein sources are described. It is important to remember that the capture process is comprised of seven major steps: (1) loading the suitable matrix into an expanded-bed column, (2) expanding the adsorbent matrix in the proper buffer, (3) upward loading of the protein sample while the matrix is in the expansion mode, (4) washing the expanded matrix in an upward direction, (5) packing the adsorbent matrix, (6) washing the packed matrix in a downward direction, and (7) eluting the protein from the packed column.
440
JOSEPH SHILOACH AND ROBERT M. KENNEDY
T A B L E 3 S T R E A M L I N E Matrices Available for Affinity Expanded-Bed Adsorption Matrix
Chelating
Heparin
r-ProteIn A
Ligand Ligand density Particle size (fjum) Mean particle size (/im) Approx. mean density (gr/mL) Expansion at 300 c m / h r pH stability Long term Short term Working velocity (cm/hr) Capacity (mg/mL) Total Dynamic
Iminodiacetic 40 fimol Cu^+/mL 100-300 200 1.2 2-3
Heparin 4 mg/ml 100-300 200 1.2 2-3
r-Protein A 6 mg/ml 8-165 120 1.3 2-3
3-13 2-14 300
4-12 4-13 300
3-10 2-11 300-500 50 m g / m L 20 m g / m L
Heparin is therapeutic grade, derived from porcine sources. Recombinant protein A expressed in Escherichia coli. Long term refers to the pH interval where the gel is stable over a long period of time without adverse effects on its chromatographic performance. Short term refers to the pH interval suitable for regeneration and cleaning. Breakthrough capacity for STREAMLINE r-Protein A in packed bed and expanded bed, at a flow velocity of 300 cm/hr. Sedimented bed height is 15 cm. Capacity for STREAMLINE Chelating and STREAMLINE Heparin is comparable ( + 1 0 % ) to the Fast Flow versions of these matrices.
A. Capturing Extracellular Proteins i. General Considerations This category includes proteins that are produced and secreted by microorganisms, prokaryotes and eukaryotes, and by mammaUan cells grown in suspension in bioreactors. The starting solution contains the targeted protein, other proteins, intact producing cells, and media components such as carbon sources, nitrogen sources, phosphate, various metals, amino acids, and antifoam. The producing cells can be bacteria, which are approximately 1 ixm in size and have a cell wall; they can be yeast, which are of the order of 10 /xm and also have a cell wall; or they can be mammaUan cells, which are also around 10 ^tm but do not have a cell wall. Mammalian cells can be easily disrupted, thus adding intracellular components such as nucleic acids to the solution, possibly interfering in the expanded-bed process. The biomass concentration of the loading solution should be between 5 and 8% (50 to 80 g packed cells per liter) or less, and the viscosity should be below 10 mPa. The loading linear flow rate should be at least 300 cm/hr. In addition, it is important to consider the binding conditions of the targeted protein to the matrix. For example, in the case of an ion-exchanger, it is important to adjust the pH and the ionic strength of the solution, sometimes requiring a dilution
441
EXPANDED-BED ADSORPTION
that can significantly reduce the protein concentration and increase the volume. As was indicated in the introduction section, a stable expanded bed is critical to the overall expanded-bed operation. However, when dealing with a solution of intact cells, cell solid concentration and ionic strength have a minimal effect on bed stability, and therefore, the entire solution can be pumped directly on the expanded matrix. It is important to mix the loading solution continuously throughout the loading process to ensure homogeneous loading. Sometimes, depending on the cell type and media components, the upper flow adaptor can be clogged, likely due to cell aggregation. When clogging occurs, it is essential to reverse the direction of the flow for a few seconds. Various expanded-bed processes differ in the protein source, the composition of the protein solution, the type of adsorbent material, and the various buffers used for the process. ii. Practical Examples
This segment includes a detailed example of capture and recovery of extracellular mutant diphtheria toxin from Cory neb acterium diphtheria^ and some key parameters and conditions for obtaining other products. a. Niutant Diphtheria Toxin from Corynebacterium diphtheriae
CRM 9 is a mutant diphtheria toxin with a molecular weight of 60,000 secreted extracellularly from C. diphtheriae under certain growth conditions.^ Traditionally, this protein is captured on an anion-exchanger only after centrifugation, clarification, and extensive diafiltration, since the ionic strength of the medium is too high for direct adsorption (Fig. 6). When using the
Fermentation ]
^ ppt -*
, 401
Centrifugation
,, Sup ^ ppt -
Clarification
,, 401 Ultrafiltration
' 30 liters 20mM Tris PH=7.4 (NH4)2 SO4
(60 OD 600nm)
2-31
Diafiltration
'
21
(sharplessAS-16P) 1 hour (sharplessAS-16P) 1 hour (10ft2 10000 MWCo) 3-4 hours (10ft2 10,000 MWCo) 2 hours
Ammonium sulfate precipitation F I G U R E 6 Flow diagram of the traditional capture process for CRM 9. (Reprinted from Shiloach and Kaufman^ by courtesy of Marcel Dekker, Inc.)
442
JOSEPH SHILOACH AND ROBERT M. KENNEDY
10rFlow rate: Loading 400 cm/hr Elution 100 cm/hr Protein loading: 40mg/ml packed bed
o 00 CVJ
<
140
280
420
liters FIGURE 7 Chromatography of C diphtheria fermentation broth on expanded bed. Sixty liters of bacterial culture, diluted six times, were applied on 4 liters STREAMLINE DEAE in a STREAMLINE 200 column. After loading, the CRM 9 was eluted using 0.175 M KCI.
expanded-bed approach, it is possible to adsorb the protein directly from the bacterial suspension only after its proper dilution (six- to eightfold).^ The expanded-bed procedure, using STREAMLINE DEAE as the adsorbent, is performed as follows: At the end of the fermentation, the culture's conductivity is adjusted to 3.5 mS/cm by diluting the culture with 20 m M Tris buffer pH 9.0 (approximately sixfold). The culture is then pumped upwards on equilibrated STREAMLINE DEAE expanded in a STREAMLINE column at a ratio of 1 liter original culture to 80-100 mL packed resin at a flow rate of 300-400 cm/hr. After all the culture is applied to the column, it is washed with the same Tris buffer until the eluent OD (optical density) at 280 nm returns to baseline. At this point, the upward flow is stopped, the bed is
T A B L E 4 Capture-Step Comparison for Processing 40 Liters of C diphtheria Culture for Production of C R M 9 Diphtheria Toxin "Traditional" process CRM 9 yield (g) Filtration area (m^) Column volume (liters) Processing time (hr) Product volume (liters)
3.3 0.97 NA 8.0 2.5
Expanded-bed process 3.0 NA 4.8 4.0 6.0
443
EXPANDED-BED ADSORPTION
Fermentation Tris buffer 40mM PH=9.0
••
''
Dilution
loading 125 I/hour [400 cm/hr] 3 hours Washing 60! 1/2 hour
40 1 (60 OD 600nm)
'
3201
*
Washing Elution 0.175M
4.81 Expanded bed
31 |/hour [100 cm/hr.] KCI. 30 min.
ii
6.0 Liters (NH4)2 SC>4 Ammonium sulfate precipitation F I G U R E 8 Flow diagram of the expanded-bed capture process for CRM 9. (Reprinted from Shiloach and Kaufman® by courtesy of Marcel Dekker, Inc.)
allowed to settle and the flow adaptor is lowered to the surface of the packed bed. The column is then washed with two column volumes of the 20 m M Tris buffer in a downward mode at a rate of 100 cm/hr, and the CRM 9 diphtheria toxin is eluted with 20 m M Tris buffer, pH 7.4 containing 0.175 M KCI. The OD peak at 280 nm is collected (Fig. 7) and the protein is precipitated with ammonium sulfate. The overall operation can be seen in Fig. 8. A comparison between the traditional process and the expanded-bed process can be seen in Table 4. b. Aprotonin from Hansenula polymorpha^
Adsorbent: STREAMLINE SP in 20 m M sodium acetate pH 3.5 Capture conditions: cell concentration 5 %, conductivity 25 mS/cm, pH3.5 Flow rate: 300 c m / h r Wash: 20 m M Sodium acetate pH 3.5 Elution: contaminants in 0.5 M NaCl in the sodium acetate buffer; aprotinin in 0.9 M NaCl. c. Recombinant Human Serum Albumin from Pichia pastoris^^
Adsorbent: STREAMLINE SP in 50 m M acetate buffer pH 4.5, containing 50 m M NaCl Capture conditions: heat-treated yeast culture, diluted 1:2, pH 4.5, information on percent solids is not available (the estimated concentration is around 20%)
444
JOSEPH SHILOACH AND ROBERT M. KENNEDY
Flow rate: 100-250 cm / h r Wash: 50 m M acetate buffer pH 4.5, containing 50 m M NaCl Elation: 100 m M phosphate buffer pH 9.0, containing 300 m M NaCl downward at 100 c m / h r d. Recombinant Human Nerve Growth Factor (rhNGF) from Cells^^
CHO
Absorbent: STREAMLINE SP in 25 m M MES/NaMES, 0.3 M sodium acetate pH 6.0 Capture conditions: temp. 37°C. Flow rate: 375 c m / h r Wash: 25 m M MES-NaMES, 0.3 M sodium acetate pH 6.0 Elution: 25 m M MES-NaMES 1 M sodium acetate pH 6.0. e. Recombinant Monoclonal Antibody from CHO Cell Culture^^
Adsorbent: STREAMLINE SP, 25 m M MES buffer pH 5.4 Capture conditions: the cell culture is kept for 2 to 3 days to settle the cells, then the supernatant, containing 5 X 10"^ cells/mL, is diluted 3 times and the pH adjusted to 5.4 Flow rate: 135-144 c m / h r Wash: 24 volumes of 25 m M MES buffer pH 5.4 Elution: linear gradient from 40 to 400 m M NaCl in 25 m M MES, pH 5.4 f. Humanized lgG4 Monoclonal Antibody from Myeloma Cell Cu/ture'^
Adsorbent: STREAMLINE Protein A in 50 m M glycine glycinate pH 8.0, containing 250 m M NaCl Capture conditions: the cell culture suspension is applied directly at 37°C. Flow rate: 300 c m / h r Wash: 50 m M glycine glycinate pH 8.0, containing 250 m M NaCl, and additional 10 column volumes, after the OD comes back to baseline Elution: 0.1 M glycine pH 3.0. iii. Discussion
As indicated in the previous section, the expanded-bed capture process for extracellular proteins from various sources is quite straightforward. Extracellular proteins from bacteria, mammalian cells and yeast can be captured, using different types of adsorbent materials, by directly pumping the cell suspension on the expanded-bed column. Preparations prior to the adsorbing process are minimal. For example, in the case of the mutant diphtheria toxin, culture dilution was needed; in the case of recombinant human serum albumin from P. pastor is ^ the culture was heat-treated; and in the case of monoclonal antibody from CHO cell, the culture was allowed to settle for 2 - 3 days before pumping it on to the anion-exchanger. Compared with the traditional approach (Fig. 1), the expanded-bed method is shorter and requires fewer steps (Table 4). The elimination of centrifugation, filtration, and ultrafiltration is likely to reduce processing time and capital ex-
EXPANDED-BED ADSORPTION
1 ^ 1
445
T A B L E 5 Economic Considerations for Obtaining C R M 9 Diphtheria Toxin"
Equipment Consumables Time (h)
Traditional
Expanded-bed adsorption
$90,000-$120,000^ $3,000-$4,300 (filters) 15 (set up and cleaning 10 hr, operation 5 hr)
$20,000'' $3,600 (resin) 14 (set up and cleaning 6 hr. operation 5 - 7 hr)
The comparison is for processing 70 liters of bacterial culture. Equipment needed: continuous centrifuge and tangential flow filtration system. Equipment needed: expanded bed column for 4.5 liter resin.
penses. Table 5 summarizes the economical considerations of the two approaches when mutant diphtheria toxin is recovered. B. Capture of Intracellular Proteins i. General Considerations
This category includes proteins that are produced inside microorganisms or mammaUan cells. To release these proteins, the biomass is first concentrated by conventional continuous centrifugation, and then it is disrupted. Therefore, the starting protein solution contains the targeted protein, other proteins, cell contents such as cell wall fragments and DNA, but it does not contain any media components. Unlike the solution that contains extracellular proteins, this solution is not homogenous, possibly affecting the bed stability; therefore, it is more difficult to achieve a stable expanded bed when dealing with intracellular proteins than when dealing with extracellular proteins. As previously stated, the bed stability is affected by the solution's ionic strength, pH, solids concentration and viscosity, and, in this case, also by the method of cell disruption. The disrupted cell suspension, after proper adjustment, is pumped directly on the expanded bed. The solid concentration in the suspension should be between 50 and 80 g per liter, the viscosity below 10 mPa, and the loading linear flow rate not less than 300 cm/hr. Unlike the extracellular protein case that deals with intact cells, in the disrupted cells case, the ionic strength has an effect on column stability. It is important to mix the loading solution continuously throughout the loading process to ensure homogeneous loading, and to adjust the binding conditions based on the properties of the targeted protein and the choice of matrix. ii. Practical Examples
This segment includes a detailed example of capture and recovery of intracellular protein (recombinant exotoxin A from E. coli) and key parameters and conditions for obtaining other products. a. Recombinant Pseudomonas aeruginosa Exotoxin A from £. co/i'^' '^
The modified P. aeruginosa exotoxin A is overproduced in £. coli BL21 (ADE3). The modified toxin, which lacks the enzymatic activity but retains the binding activity, is missing glutamic acid in position 553. The expression
446
JOSEPH SHILOACH AND ROBERT M. KENNEDY
4.5 kg Bacteria Sucrose20 mM Buffer90 I Suspension Centrifugation
2-3 h
Clarification
2-311
DEAE Sepharose fast flow 25.5 1 30 cm X 36 cm
loading 1h elution 2h .(100 cm h""")
I 36 1 F I G U R E 9 Flow diagram of the traditional capture process of recombinant exotoxin A from £ coli. (Reprinted from Johansson et o/.'^ with permission from Elsevier Science.)
of the modified toxin is controlled by the T7 promoter and the protein accumulates in the periplasmic space following induction with IPTG. When applying the traditional recovery process (Fig. 9), at the end of the fermentation, the cells are collected and suspended in two volumes of 20% sucrose in 20 m M Tris buffer pH 7.4 containing 1 m M EDTA. After the cells are well dispersed, they are mixed for an additional 10 min and then diluted with 18 volumes of 20 m M Tris buffer pH 7.4. The cell suspension is centrifuged and microfiltered through a 0.45 jubm filter. The clarified supernatant is applied on DEAE Sepharose Fast Flow at a flow rate of 100 c m / h r and a ratio of 40 OD280 nm per 1 mL. column. The elution is performed with a linear gradient of eight column volumes from 0 to 0.5 M NaCl in the Tris buffer. In contrast, when applying the expanded-bed process (Fig. 10), the collected cells are suspended in 2 volumes of 20% sucrose in 50 m M Tris buffer pH 7.4 containing 1 m M EDTA. After the cells are well dispersed.
4.5 kg E Bacteria Sucrose-
•
50 mM buffer180ISu spension
.
.
(400 cm h ') 2 1 min ^ loading-1801 washing - 40 1 2h
i Streamline DEAE 4.71 20 cm X 15 cm
• 0.5 NaCI 0.5 I min""" (100 cm h-'')
t ,I \
13 51 F I G U R E 10 Flow diagram of the expanded-bed capture process of recombinant exotoxin A from £. coli. (Reprinted from Johansson et 0/.'^ with permission from Elsevier Science.)
EXPANDED-BED ADSORPTION m
447
T A B L E 6 Capture Step Comparison for Processing 4.5 Kg £. Co/f Cells for Production of Exotoxin A 553D a
Packed-bed DEAE Sepharose FF Specific activity after the step (mg toxin/mg protein) Recovery (%) Processing time (hr) Column volume (liters) Eluent volume (liters)
0.1 73.0 8.0 25.5 36.0
Expanded-bed STREAMLINE DEAE 0.06 79.0 2.5 4.7 13.5
Reprinted from Johansson et al}^ W\t\v permission from Elsevier Science. Packed column numbers are extrapolated values.
they are mixed for an additional 10 min and then are diluted with 18 volumes of 50 m M Tris buffer, pH 7.4. The mixture is mixed for an additional 10 min, endonuclease (Benzonase) is added at a ratio of 7S units per gram cells, and the suspension is diluted with 18 volumes of the Tris buffer. The cell suspension is then pumped upward on an equilibrated and expanded STREAMLINE DEAE at a ratio of 1 g cell extract per 1 mL packed column at a flow rate of 400 cm/hr. After all the cell suspension is applied on the column, the column is washed with 50 m M Tris buffer pH 7.4 until the OD at 280 m M of the eluent is back to baseline. At this point, the upward flow is stopped, the bed is allowed to settle, and the flow adaptor is lowered to the surface of the packed bed. After washing in a downward mode with 2 column volumes at a rate of 100 cm/hr, the protein is eluted with 20 m M Tris buffer pH 7.4 containing 0.5 M NaCl. The OD280 P^ak is collected for the next purification step. Successful adsorption procedure is achieved using the following conditions: concentration of the dry cell mass is 6 g/liter, the buffer concentration is 50 m M and the endonuclease content is 75 units per gram cells. The comparison between the traditional process and the expended-bed process is summarized in Table 6. b. Recombinant
Diphtheria
Toxin from £. c o / i ' *
Method of cell disruption: £. colt cells are suspended in the binding buffer (0.25 gram cells per mL), and the suspension is passed twice through a homogenizer (Manton Gaulin) at 9000 psi. Absorbent: STREAMLINE Chelating, saturated with NiS04 and equiUbrated in 20 m M Tris, pH 8.0 containing 0.5 M NaCl and 5 m M imidazole. (The protein is expressed with His-tag). Capturing conditions: broken cell suspension at a concentration of 0.13 g / m L in binding buffer. Flow rate: 200 m L / h Wash: 20 m M Tris pH 7.2 containing 0.5 M NaCl and 30 m M imidazole.
448
JOSEPH SHILOACH AND ROBERT M. KENNEDY
Elution: 20 m M Tris pH 7.2, containing 0.5 M NaCl and 250 m M imidazole c. Recombinant Annexin V from £. coli^^
Method of cell disruption: cell suspension is passed three times through a high-pressure homogenizer 10,500-13,500 psig. Adsorbent: STREAMLINE DEAE equilibrated in 30 m M ammonium acetate pH 5.5 Capturing conditions: broken cell suspension at a concentration of 3.6% solids in 30 m M ammonium acetate pH 5.5; the suspension contains 1% detergent Triton X-100 Flow rate: 300 c m / h r Wash: 30 m M ammonium acetate pH 5.5 Elution: 30 m M ammonium acetate pH 5.5 containing 250 m M NaCl 100 c m / h r d. Recombinant Anti-HIV Fab-Fragment from £. co/i'®
Method of cell disruption: cell suspension passed three times through high-pressure homogenizer at 10,000 psig Adsorbent: STREAMLINE SP in 50 m M sodium acetate pH 5.0 Capturing conditions: dry cell concentration, 1.4% suspended in 50 m M sodium acetate pH 5.0, containing endonuclease (benzonase) at a ratio of 10 fiL per 60 g cells (dry weight). Flow rate: 300 c m / h r Washing: 50 m M sodium acetate pH 5.0 Elution: 50 m M sodium acetate pH 5.0, containing 1 M NaCl, at 100 cm/hr e. Recombinant lnterluldn-8 from £. co/i Inclusion Bodies
Method of cell disruption: cell pellet suspended in three volumes of 30 m M sodium phosphate pH 6.5 containing 6 M guanidine hydrochloride. Suspension is mixed for 3 hr and diluted with six volumes of water in two steps: three volumes in the first step followed by mixing for 30 min, and three volumes in the second step followed by mixing overnight. Adsorbent: STREAMLINE SP in 30 m M sodium phosphate pH 6.5 Capturing conditions: 1% (dry weight) cell suspension in 30 m M sodium phosphate containing 6 M guanidine hydrochloride Flow rate: 300 c m / h r Washing: 30 m M sodium phosphate containing 6 M guanidine hydrochloride Elution: 30 m M sodium phosphate containing 6 M guanidine hydrochloride, and 0.5 M NaCl at a flow rate of 100 c m / h r f. Glucose 6 Phosphate Dehydrogenase (G6PDH) from S. cerevisiae^^
Method of cell disruption: yeast suspension 50% wet weight is passed through a bead mill using 0.5 mm glass beads
449
EXPANDED-BED ADSORPTION
Adsorbent: STREAMLINE DEAE in 50 m M sodium phosphate pH 6.0 Capturing conditions: 6.5% (dry weight) yeast suspension in 50 m M sodium phosphate pH 6.0, viscosity 5.0 mPa Flow rate: 200 c m / h r Washing: 2 5 % glycerol in 50 m M sodium phosphate pH 6.0 in an expanded mode, followed by 50 m M NaCl in 50 m M sodium phosphate pH 6.0 in a packed mode Elution: 150 m M NaCl in 50 m M sodium phosphate pH 6.0 iii. Discussion
Capturing intracellular proteins on an expanded bed involves preliminary steps that are not required for the capture of extracellular proteins. These steps are the initial cell concentration, usually by filtration or centrifugation, and cell disruption, using a high-pressure homogenizer, a bead mill, or osmotic shock. To prepare a disrupted cell suspension for direct adsorption on an expanded bed, suspension viscosity and ionic strength can be adjusted by adding detergent, endonuclease, and salt. It is also important to adjust the concentration of solids in the broken cell suspension because it affects the stability of the expanded matrix. As seen in the example of capturing recombinant Interlukin-8 from £. coli, the expanded-bed process is also suitable for capturing a protein extracted from inclusion bodies using guanidine hydrochloride. Since concentration of the cells is an essential preliminary step when processing intracellular proteins, and instrumentation for this unit operation is required, a comparson of the economical considerations between the described approach and the traditional method is different from the case of the extracellular proteins (Table 7).
Yll. DISCUSSION AND CONCLUSIONS The production examples described here, and the information published since this chapter was written, clearly indicate that the expanded-bed operation offers an efficient alternative to the conventional protein capturing process.
T A B L E 7 Economic Considerations for Obtaining Recombinant Exotoxin A (533D)°
Equipment Consumables Time (hr)
Traditional
Expanded-bed adsorption
$103,000 $1,500 (filters) $10,000 (resin) 20 (set up and cleaning 11 hr. operation 8-9 hr)
$90,000"" $3,600 (resin) 12.5 (set up and cleaning 6.5 hr. operation 6 hr)
The comparison is for processing 4.5 kg of bacterial biomass. Equipment needed: continuous centrifuge and chromatography column. Equipment needed: continuous centrifuged and expanded-bed column for 4.5 liter
450
JOSEPH SHILOACH AND ROBERT M. KENNEDY
The latter requires that the starting product solution be centrifuged, clarified, and at times, dialyzed, before it can be adsorbed on a packed column. In contrast, capturing the desired product by the expanded-bed method is simpler because the crude starting solution is pumped directly on the adsorbent. To achieve a successful expanded-bed process operation, the adsorbent must be stable in its expanded mode, requiring that careful evaluation of the loading conditions take place during the process development stage. The follow^ing points should be considered: establishing optimal loading conditions for intracellular proteins was found to be more complicated than for extracellular proteins; starting solutions with high viscosity and solid content often must be diluted; and when the loading solution is very crude, the upper flow adaptor may get clogged, requiring that the flow direction be changed frequently. Note that the expanded-bed approach is not a viable option if the binding of the desired product to the adsorbent is possible only after initial partial purification. In summary, when evaluating the suitability of the expanded-bed approach for capturing a particular protein, one must consider whether the product is from bacteria, yeast, or mammalian cells; whether it is intracellular or extracellular; and the specific activity of the final product, its overall yield, the processing time and capital, labor, and maintenance costs.
REFERENCES 1. Chase, H. A. (1994). Purification of proteins by adsorption chromatography in expanded beds. Trends Biotechnol. 12, 296-303. 2. Hjorth, R. (1997). Expanded bed adsorption in industrial bioprocessing: Recent developments. Trends Biotechnol. 15, 230-235. 3. Gailliot, E. P., Gleason, C , Wilson, J., and Zwarick, J. (1990). Fluidized bed adsorption for whole broth extraction. Biotechnol. Prog. 6, 370-375. 4. Thommes, J. (1997). Fluidized bed adsorption as a primary recovery step in protein purification. Adv. Biochem. Eng. Biotechnol. 58, 185-230. 5. Amersham Pharmacia (1995). "Application Note: Cleaning in Place," Publ. No. 18-1115-27. 6. Hanson, K. A. (1995). Physical chemical properties of STREAMLINE Ion exchangers. Poster Presentation, Eur. Congr. Biotechnol., 7th, Nice, France. 7. Fass, R., Bahar, S., Kaufman, J., and Shiloach, J. (1995). High yield production of diphtheria toxin mutants by high density culture of C7(6)*'^^^ strains grov^n in a non deferrated-media. Appl. Microbiol. Biotechnol. 43, 83-88. 8. Shiloach, J., and Kaufman, J. B. (1999). The combined use of expanded bed adsorption and gradient elution for capture and partial purification of mutant diphtheria toxin (CRM 9) from Cory neb acterium diphtheriae. Sep. Sci. Technol. 34, 29-40. 9. Zurek, C , Kubis, E., Keup, P., Hoerlein, D., Beunink, J., Thoemmes, J., Kula, M. R., Hollenberg, C. P., and Gellissen, G. (1996). Production of two Aprotonin variants in Hansenula plymorpha. Process Biochem. 31, 679-689. 10. Noda, M., Sumi, A., Ohumura, T., and Yokoyama, K. (1966). Process for purifying recombinant human serum albumin. Eur. Pat. Appl. EP 0 699 687 A2. 11. Beck, J., Liten, A., Viswanathan, S., Emery, C , and Builder, S. (1996). Direct capture of Nerve Grov^th Factor from CHO cell culture by EBA. Presentation, Recovery Biol. Prod., 8th, Tucson, AZ. 12. Batt, B. v., Yabannavar, V. M., and Singh V. (1995). Expanded bed adsorption process for protein recovery from whole mammalian cell culture broth. Bioseparation 5, 4 2 - 5 3 .
EXPANDED-BED ADSORPTION
45 I
13. Jagerstern, C , Johansson, S., Bonnerjea, J., and Pardon, R. (1996), Capture of a humanized IgG4 directly from the fermentor using STREAMLINE r-protein A. Presentation, Recovery of Biol. Prod., 8th, Tucson, AZ. 14. Fass, R., Van de Walk, M., Shiloach, A., Joslyn, A., Kaufman, J., and Shiloach, J. (1991). Use of high-density cultures of Escherichia coli for high level production of recombinant Pseudomonas aeruginosa exotoxin A. Appl. Microbiol. Biotechnol. 36, 65-69. 15. Johansson, H. J., Jagersten, C , and Shiloach, J. (1996). Large-scale recovery and purification of periplasmic recombinant protein from E. coli using expanded bed adsorption chromatography follow^ed by new ion exchange media. / . Biotechnol. 48, 9-14. 16. Noronha, S., Kaufman, J., and Shiloach, J. (1999). Use of STREAMLINE Chelating for capture and purification of poly His Tagged recombinant proteins. Bioseparation 8, 145-151. 17. Barnfield Frej, A. K., Hjorth, R., and Hammarstsrom A. (1994). Pilot scale recovery of recombinant Annexin V from unclarified £. coli homogenate using expanded bed adsorption. Biotechnol. Bioeng. 44, 922-929. 18. Jagerstern, C. (1994). Purification of recombinant Anti-HIV Fab-fragment expressed in £. coli. Presentation, Recovery Biol. Prod., 7th San Diego, CA. 19. Chang, Y. K., and Chase, H. A. (1966). Ion-exchange purification of G6PDH from unclarified yeast cell homogenate using expanded bed adsorption. Biotechnol. Bioeng. 49, 204-216.
ROBERT M. K E N N E D Y Separations Group, Amersham Pharmacia Biotech, Piscataway, New Jersey 088SS
I. II. III. IV. V. VI.
INTRODUCTION PRINCIPLES OF EXPANDED-BED OPERATION EXPERIMENTAL STRATEGY INSTRUMENTATION MATRICES APPUCATIONS A. Capturing Extracellular Proteins B. Capture of Intracellular Proteins VII. DISCUSSION AND CONCLUSIONS REFERENCES
INTRODUCTION
The complete process for obtaining pure proteins can be divided into three main steps: capture, purification, and poHshing. The first step is the immobihzation of the target protein onto some adsorptive surface, and it can be viewed as a combination of clarification, concentration, stabilization and initial purification. Because the starting protein solution (feedstock) is usually crude, it is essential to clarify the solution. The traditional or conventional approach involves centrifugation, microfiltration, ultrafiltration, or diafiltration before the target protein solution can be loaded on an adsorbing material, utilizing packed-bed chromatography. The clarification step is a demanding operation, and is particularly difficult v^hen processing large quantities of microorganisms, especially disrupted microorganisms. Highspeed, large-scale centrifugation and microfiltration are the most common processes used to obtain protein solutions that are suitable for packed-bed chromatography; therefore, it is obvious that an approach that eliminates the Separation Science and Technology, Volume 2 Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved.
43 I
432
JOSEPH SHILOACH AND ROBERT M. KENNEDY
clarification step can significantly simplify and improve the overall purification process. Direct adsorption of the protein not only eliminates the clarification step, but also produces a concentrated and partially purified product ready for the next purification step (Fig. 1). Several protein capture procedures, such as batch adsorption, solvent extraction, and expanded-bed adsorption, are protein capture procedures that do not require centrifugation and filtration. In this chapter wt describe the expanded-bed adsorption approach for capturing target proteins. In the expanded-bed mode, the starting protein solution is pumped upw^ard through a bed of adsorbent beads that are constrained by a flow^ adaptor.^ As a result of the upward flow and the properties of the beads, the bed expands as spaces open up between the beads. If the physical properties of the beads are significantly different from those of the particles in the feedstock, the particles can pass freely through the bed without being trapped.
Extracellular
Fermentation
Intracellular
Cell separation ppt (Cells)
Sup.
Clarification
Suspension
Concentration
Cell disruption
Dialysis
Clarification
Column chromatography
Concentration
f
' Dialysis
Column chromatography
.--
F I G U R E I General purification schemes of extracellular and intracellular proteins using traditional and expanded-bed approaches. The solid lines indicate the traditional approach and the broken lines the expanded-bed approach. Using the expanded bed, when the product is extracellular, the fermentation broth Is pumped directly on the column; when the product is intracellular, the suspension of disrupted cells is pumped directly on the column.
433
EXPANDED-BED ADSORPTION
An effective process, which means the formation of a stable expanded bed, depends on parameters such as the viscosity, ionic strength, sohds content, and pH of the feedstock as well as the linear flow rate. Recently, the expanded-bed approach has been successfully used to combine clarification, concentration, and initial purification. The references cited in two recent reviews provide a good summary of these applications.^'^
II. PRINCIPLES OF EXPANDED-BED OPERATION The expanded bed is, in principle, similar to a fluidized bed, a common unit-operation in the chemical industry.^ However, in the expanded-bed method, mixing of the adsorbent material is minimal during the operation, whereas a fluidized bed is turbulent. This means that an expanded bed is more effective for adsorption and elution processes than the fluidized bed. A complete review of all engineering aspects, comparing expanded beds and fluidized beds, can be found in Thommes."^ The expanded-bed adsorption operation is illustrated in Fig. 2. The overall operation is comprised of several phases. In the first phase, the adsorbent material is expanded and equiUbrated by applying an upward liquid flow to the column. To allow for sufficient contact time and efficient binding of the target molecule, the expansion should be three times that of the sedimented bed ("expansion ratio"), to a height of approximately 50 cm. A stable bed is formed when the adsorbent particles achieve equilibrium between particle sedimentation velocity and upward liquid flow velocity. In the second phase, the sample is applied to the expanded adsorbent. The crude, unclarified protein solution of intact or disrupted biomass is pumped upward on the column. In a well-defined process, the expanded adsorbent will remain stable and will not change its expansion ratio. However, if the
Before start-up sedimented adsorbent FIGURE 2
Expansion and equilibration of the adsorbent
Application of feed, followed by washing
Schematic presentation of the steps of expanded-bed adsorption.
Elution in packed bed
434
JOSEPH SHILOACH AND ROBERT M. KENNEDY
process is not defined, there is a chance that the expanded adsorbent will not remain stable and will collapse or expand to the top of the column. More details are provided later in Section VI, "Application." During loading, the target proteins are bound to the adsorbent matrix, while cell debris and contaminants pass through the bed and the upper flow adaptor. The third phase is the washing of the expanded adsorbent (in an upward flow) using a buffer, most often the loading buffer. During this step, weakly bound materials, such as residual cells, cell debris, and other types of particles, are washed out. Following this step, the flow through the column is stopped and the adsorbent material settles. The upper flow adaptor is lowered to the surface of the settled adsorbent and the flow is reversed, going downward. The captured proteins are eluted from the sedimented bed, using appropriate elution solutions. The final phase, as in other adsorption processes, is the washing and regeneration of the adsorbing material. Detailed information is provided by the manufacturer.^ It is possible to use the adsorbent for a large number of runs without losing its functionality. As indicated, expanded-bed adsorption is based on controlled, stable fluidization, thus combining the hydrodynamic properties of a fluidized bed with the chromatographic properties of a packed bed. Stable fluidization with minimal backmixing results in a performance that is similar to that of a traditional plug flow column. The stability of the bed arises from the special design of the distributor plate at the base of the column and from two properties of the adsorbent: particle size distribution and particle density distribution. The distributor plate generates the required pressure drop and directs the flow only in a vertical direction, thus eliminating any radial flow that might create turbulence through the column. The size distribution and the density distribution of the beads ensure minimal local mixing. Table 1 summarizes the work of Hanson^ and illustrates the distribution of particle size and density in the bed. Particle distribution is described from the top of the column down: the top 30 mL contain 12% of the particles with an average size of 144 ^tm and an average density of 1.15 gm/mL, while the bottom 104 mL, 4 1 % of the expanded bed volume, contain particles with an average size of 238 fim and an average density of 1.19 g/mL. The polydispersity of the beads positions them at specific heights in the expanded bed; the smaller, lighter particles move to positions at the top and the larger.
TABLE I
Distribution of Beads in an Expanded Bed°
Sedimented gel volume (mL)
%
Average particle size ()Jim)
Density (gm / mL)
30 49 72 104
12 19 28 41
144 164 186 238
1.15 1.16 1.17 1.19
Degree of expansion: 2.5 folds in water, at 300 cm/hr, in a 5.0 cm diameter column.
435
EXPANDED-BED ADSORPTION
o o
o
o o o
Upper adapter o o o o O
no o
o , '^o
Qp^ O O ^
Lower adapter
booOocn I Flow
FIGURE 3
Packed-bed (a) and expanded-bed (b) columns.
heavier particles move to the bottom, resulting in a stable expansion (Fig. 3). In other v^ords, the beads find their ideal position in the column, v^hich is the reason for the lowr axial dispersion in the expanded-bed operation. The high density of the beads is necessary in order to be able to run the expanded bed at high flow velocities, thus achieving higher productivity.
III. EXPERIMENTAL STRATEGY As with other purification procedures, to develop an expanded-bed procedure, one must first follow an experimental strategy composed of method scouting, method optimization, process verification, and production. The purpose of method scouting is to define the most suitable adsorbent and the optimal conditions for binding and elution. The adsorbent of choice should be the one showing the strongest binding of the target protein, while hardly binding contaminating proteins, demonstrating the highest selectivity a n d / o r the highest capacity for the protein of interest. These initial determinations are done using packed-bed chromatography with clarified feedstock. The conditions must be verified and adjusted using the expanded bed with unclarified feedstock. The feedstock flows through the expanded bed at the relatively high flow rate of 300 to 400 cm/hr; therefore, it is important to ensure that the binding parameters of the expanded bed with the unclarified feedstock are similar to those determined using the packed bed process. Figure 4 shows breakthrough capacity curves of three different columns, one packed-bed and two expanded-bed that represent laboratory and pilot scale productions. Increasing amounts of BSA were loaded into the columns and the fraction of the unbound BSA was monitored. Please note the similar
436
JOSEPH SHILOACH AND ROBERT M. KENNEDY
C/Co Packed XK16
1.0
Streamline® 50 Streamline 200
0.8
0.6
h
0.4
0.2
0.0 00
20
40
60
Applied BSA (mg/ml adsorbent) FIGURE 4 Breakthrough curves for BSA on STREAMLINE DEAE. The breakthrough capacity is determined using frontal analysis.
shapes of the breakthrough curves for the packed-bed and expanded-bed modes, as well as for the different sizes of expanded-bed columns with internal diameters of 50 and 200 mm. This similarity demonstrates that the adsorption performance is the same in expanded-bed and packed-bed modes. The special expanded-bed column (STREAMLINE, Amersham Pharmacia Biotech, Sweden) can be scaled up while maintaining the same performance. Once a suitable maximum loading capacity is defined, it is suggested that a safety margin of 75% of the breakthrough capacity be applied, to compensate for sample variability that can affect the binding capacity. The translation of the packed-bed conditions to expanded-bed operation is done with crude, unclarified feedstock and an expanded-bed column with a diameter of 2.5 or 5 cm. Since proper operation of the expanded bed depends on stable expanded adsorbent, and the stability of the adsorbent depends on the viscosity, the ionic strength and the flow velocity of the loading solution, these parameters are adjusted.
437
EXPANDED-BED ADSORPTION
Cleaning-in-place (CIP) procedures^ are important for expanded-bed operations due to the high debris load. Most other chromatographic matrices are never exposed to such levels of contamination. It is also important to verify that bed characteristics have not been altered after the CIP. After establishing process parameters, pilot-scale work may begin with a pilot-scale column (e.g., 20 cm in diameter). For some applications, this column may be suitable for the final production, since such a column has a capacity for 4.7 liters of matrix. Further scale-up is accomplished by increasing the column diameter while maintaining the sedimented bed height, the flow velocity, and the expanded bed height. Larger columns currently in use have a diameter of 1 m.
IV. INSTRUMENTATION The basic configuration and optional setup of the expanded-bed operation is shown in Fig. 5. In addition to the special column, there is a need for two pumps and a set of valves to control the flow to and from the column. Columns for expanded-bed adsorption are commercially available from vari-
Hydraulic liquid
Waste F I G U R E 5 Basic configuration of expanded-bed column operation. Pump 2 pushes down the flow adaptor, pump I controls the upward flow of the sample and the various buffers. Valve VI is a four-way valve directing the flow to and from the column.
438
JOSEPH SHILOACH AND ROBERT M. KENNEDY
ous sources. The two main features that characterize the STREAMLINE columns are the unique design of the liquid distribution system and a moveable adaptor. Columns range in size, from 2.5 cm in diameter used for method development, 5 and 10 cm used for process development, and up to 1200 cm used for production purposes. Columns ranging from 2.5 to 200 cm are built from glass, while the larger ones are made of stainless steel. Another company, Upfront (Denmark), offers an expanded-bed column with a special mixing device at its base. As mentioned, the expanded-bed operation requires two pumps: one for pumping the feedstock, the wash, and the elution solutions on the column, and the other for lowering the flow adaptor. To control flow direction, four-way valves and two-way valves are needed. Manual and automated control systems are available for operating an expanded-bed column.
V. MATRICES
Gel matrices for expanded-bed adsorption are commercially available, and their fluidization properties are reviewed by Thommes. 4 The adsorbent (e.g., STREAMLINE DEAE) exhibits a Gaussian-like distribution of particle sizes and particle densities. As explained previously, this particle polydispersity is an important factor that contributes to the stability of the expanded bed. The mean particle size of STREAMLINE beads is 200 ~m compared to approximately 90 ~m for matrices used in packed-bed chromatography, and the density is about 1.2 g/mL. The various matrices available under the STREAMLINE name are summarized in Tables 2 and 3. The ion-exchangers are based on highly crosslinked 6% agarose modified by including an inert quartz core to give the desired density. In the XL versions of the STREAMLINE matrices, long molecules of dextran are coupled to agarose, and the ion-exchange groups are then attached to these dextran chains. The STREAMLINE protein A is based on highly crosslinked 4% agarose modified by including an inert metal alloy core. Not mentioned in the tables is STREAMLINE phenyl used for hydrophobic interaction chromatography.
VI. APPLICATIONS
Sections I-V addressed the principles of expanded-bed operation, the instrumentation, and the special matrices. This section focuses on applications of the expanded bed, describing actual processes and providing operational guidelines. The expanded-bed adsorption process replaces the centrifugation, clarification, and concentration steps and allows the direct adsorption of the
439
EXPANDED-BED ADSORPTION
T A B L E 2 S T R E A M L I N E Matrices Available for Ion-Exchange Expanded-Bed Adsorption Matrix
SP
SPXL
DEAE
QXL
Type of ion-exchanger Ionic capacity (mmol/mL) Particle size ({xm) Mean particle size (/xm) Approx. mean density (gm/mL) Expansion at 300 c m / h r pH stability^ Long term Short term Working velocity (cm/hr) Capacity (mg/mL) Lysozyme BSA
Strong cation 0.17-0.24 100-300 200 1.2 2-3
Strong cation 0.18-0.24 100-300 200 1.2 2-3
Weak anion 0.13-0.21 100-300 200 1.2 2-3
Strong anion 0.23-0.33 100-300 200 1.2 2-3
4-13 3-14 300
4-13 3-14 300-500
2-13 2-14 300
2-12 2-14 300-500
>60 n.d.
>140 n.d.
n.d. >40
n.d. >110
Total ionic capacity by titration. Long term refers to the pH interval v^here the gel is stable over a long period of time w^ithout adverse effects on its chromatographic performance. Short term refers to the pH interval suitable for regeneration and cleaning. For the SP and the DEAE matrices, breakthrough capacity v^as determined in a STREAMLINE 50 column at a How velocity of 300 c m / h r using a 2.0 m g / m L protein solution in 50 m M sodium phosphate buffer pH 7.5 (lysozyme) and 50 m M Tris-HCl buffer pH 7.5 (BSA). Sedimented bed height w^as 15 cm. For the SPXL and QXL matrices, breakthrough capacity v^as determined in a packed bed, 4.4 mL, at a flov^ velocity of 300 c m / h r using a 2.0 m g / m L solution of protein in 50 m M glycine-HCl buffer pH 9.0 (lysozyme) and 50 m M Tris-HCl buffer pH 7.5 (BSA). Bed height was 10 cm.
protein on a proper matrix. The conditions of a specific process depend on the source and the properties of a particular protein and on its location in the microorganism. The protein can be obtained from prokaryotic and eukaryotic microorganisms (native or recombinant), from mammalian cells and tissues, insect cells, plant tissues, and other sources. In some cases, the protein can be intracellular and in others extracellular. This application section is, therefore, divided into two parts: one covering the capture of extracellular proteins and the other covering the capture of intracellular proteins. In each case, general considerations are given, and several examples of different matrices and different protein sources are described. It is important to remember that the capture process is comprised of seven major steps: (1) loading the suitable matrix into an expanded-bed column, (2) expanding the adsorbent matrix in the proper buffer, (3) upward loading of the protein sample while the matrix is in the expansion mode, (4) washing the expanded matrix in an upward direction, (5) packing the adsorbent matrix, (6) washing the packed matrix in a downward direction, and (7) eluting the protein from the packed column.
440
JOSEPH SHILOACH AND ROBERT M. KENNEDY
T A B L E 3 S T R E A M L I N E Matrices Available for Affinity Expanded-Bed Adsorption Matrix
Chelating
Heparin
r-ProteIn A
Ligand Ligand density Particle size (fjum) Mean particle size (/im) Approx. mean density (gr/mL) Expansion at 300 c m / h r pH stability Long term Short term Working velocity (cm/hr) Capacity (mg/mL) Total Dynamic
Iminodiacetic 40 fimol Cu^+/mL 100-300 200 1.2 2-3
Heparin 4 mg/ml 100-300 200 1.2 2-3
r-Protein A 6 mg/ml 8-165 120 1.3 2-3
3-13 2-14 300
4-12 4-13 300
3-10 2-11 300-500 50 m g / m L 20 m g / m L
Heparin is therapeutic grade, derived from porcine sources. Recombinant protein A expressed in Escherichia coli. Long term refers to the pH interval where the gel is stable over a long period of time without adverse effects on its chromatographic performance. Short term refers to the pH interval suitable for regeneration and cleaning. Breakthrough capacity for STREAMLINE r-Protein A in packed bed and expanded bed, at a flow velocity of 300 cm/hr. Sedimented bed height is 15 cm. Capacity for STREAMLINE Chelating and STREAMLINE Heparin is comparable ( + 1 0 % ) to the Fast Flow versions of these matrices.
A. Capturing Extracellular Proteins i. General Considerations This category includes proteins that are produced and secreted by microorganisms, prokaryotes and eukaryotes, and by mammaUan cells grown in suspension in bioreactors. The starting solution contains the targeted protein, other proteins, intact producing cells, and media components such as carbon sources, nitrogen sources, phosphate, various metals, amino acids, and antifoam. The producing cells can be bacteria, which are approximately 1 ixm in size and have a cell wall; they can be yeast, which are of the order of 10 /xm and also have a cell wall; or they can be mammaUan cells, which are also around 10 ^tm but do not have a cell wall. Mammalian cells can be easily disrupted, thus adding intracellular components such as nucleic acids to the solution, possibly interfering in the expanded-bed process. The biomass concentration of the loading solution should be between 5 and 8% (50 to 80 g packed cells per liter) or less, and the viscosity should be below 10 mPa. The loading linear flow rate should be at least 300 cm/hr. In addition, it is important to consider the binding conditions of the targeted protein to the matrix. For example, in the case of an ion-exchanger, it is important to adjust the pH and the ionic strength of the solution, sometimes requiring a dilution
441
EXPANDED-BED ADSORPTION
that can significantly reduce the protein concentration and increase the volume. As was indicated in the introduction section, a stable expanded bed is critical to the overall expanded-bed operation. However, when dealing with a solution of intact cells, cell solid concentration and ionic strength have a minimal effect on bed stability, and therefore, the entire solution can be pumped directly on the expanded matrix. It is important to mix the loading solution continuously throughout the loading process to ensure homogeneous loading. Sometimes, depending on the cell type and media components, the upper flow adaptor can be clogged, likely due to cell aggregation. When clogging occurs, it is essential to reverse the direction of the flow for a few seconds. Various expanded-bed processes differ in the protein source, the composition of the protein solution, the type of adsorbent material, and the various buffers used for the process. ii. Practical Examples
This segment includes a detailed example of capture and recovery of extracellular mutant diphtheria toxin from Cory neb acterium diphtheria^ and some key parameters and conditions for obtaining other products. a. Niutant Diphtheria Toxin from Corynebacterium diphtheriae
CRM 9 is a mutant diphtheria toxin with a molecular weight of 60,000 secreted extracellularly from C. diphtheriae under certain growth conditions.^ Traditionally, this protein is captured on an anion-exchanger only after centrifugation, clarification, and extensive diafiltration, since the ionic strength of the medium is too high for direct adsorption (Fig. 6). When using the
Fermentation ]
^ ppt -*
, 401
Centrifugation
,, Sup ^ ppt -
Clarification
,, 401 Ultrafiltration
' 30 liters 20mM Tris PH=7.4 (NH4)2 SO4
(60 OD 600nm)
2-31
Diafiltration
'
21
(sharplessAS-16P) 1 hour (sharplessAS-16P) 1 hour (10ft2 10000 MWCo) 3-4 hours (10ft2 10,000 MWCo) 2 hours
Ammonium sulfate precipitation F I G U R E 6 Flow diagram of the traditional capture process for CRM 9. (Reprinted from Shiloach and Kaufman^ by courtesy of Marcel Dekker, Inc.)
442
JOSEPH SHILOACH AND ROBERT M. KENNEDY
10rFlow rate: Loading 400 cm/hr Elution 100 cm/hr Protein loading: 40mg/ml packed bed
o 00 CVJ
<
140
280
420
liters FIGURE 7 Chromatography of C diphtheria fermentation broth on expanded bed. Sixty liters of bacterial culture, diluted six times, were applied on 4 liters STREAMLINE DEAE in a STREAMLINE 200 column. After loading, the CRM 9 was eluted using 0.175 M KCI.
expanded-bed approach, it is possible to adsorb the protein directly from the bacterial suspension only after its proper dilution (six- to eightfold).^ The expanded-bed procedure, using STREAMLINE DEAE as the adsorbent, is performed as follows: At the end of the fermentation, the culture's conductivity is adjusted to 3.5 mS/cm by diluting the culture with 20 m M Tris buffer pH 9.0 (approximately sixfold). The culture is then pumped upwards on equilibrated STREAMLINE DEAE expanded in a STREAMLINE column at a ratio of 1 liter original culture to 80-100 mL packed resin at a flow rate of 300-400 cm/hr. After all the culture is applied to the column, it is washed with the same Tris buffer until the eluent OD (optical density) at 280 nm returns to baseline. At this point, the upward flow is stopped, the bed is
T A B L E 4 Capture-Step Comparison for Processing 40 Liters of C diphtheria Culture for Production of C R M 9 Diphtheria Toxin "Traditional" process CRM 9 yield (g) Filtration area (m^) Column volume (liters) Processing time (hr) Product volume (liters)
3.3 0.97 NA 8.0 2.5
Expanded-bed process 3.0 NA 4.8 4.0 6.0
443
EXPANDED-BED ADSORPTION
Fermentation Tris buffer 40mM PH=9.0
••
''
Dilution
loading 125 I/hour [400 cm/hr] 3 hours Washing 60! 1/2 hour
40 1 (60 OD 600nm)
'
3201
*
Washing Elution 0.175M
4.81 Expanded bed
31 |/hour [100 cm/hr.] KCI. 30 min.
ii
6.0 Liters (NH4)2 SC>4 Ammonium sulfate precipitation F I G U R E 8 Flow diagram of the expanded-bed capture process for CRM 9. (Reprinted from Shiloach and Kaufman® by courtesy of Marcel Dekker, Inc.)
allowed to settle and the flow adaptor is lowered to the surface of the packed bed. The column is then washed with two column volumes of the 20 m M Tris buffer in a downward mode at a rate of 100 cm/hr, and the CRM 9 diphtheria toxin is eluted with 20 m M Tris buffer, pH 7.4 containing 0.175 M KCI. The OD peak at 280 nm is collected (Fig. 7) and the protein is precipitated with ammonium sulfate. The overall operation can be seen in Fig. 8. A comparison between the traditional process and the expanded-bed process can be seen in Table 4. b. Aprotonin from Hansenula polymorpha^
Adsorbent: STREAMLINE SP in 20 m M sodium acetate pH 3.5 Capture conditions: cell concentration 5 %, conductivity 25 mS/cm, pH3.5 Flow rate: 300 c m / h r Wash: 20 m M Sodium acetate pH 3.5 Elution: contaminants in 0.5 M NaCl in the sodium acetate buffer; aprotinin in 0.9 M NaCl. c. Recombinant Human Serum Albumin from Pichia pastoris^^
Adsorbent: STREAMLINE SP in 50 m M acetate buffer pH 4.5, containing 50 m M NaCl Capture conditions: heat-treated yeast culture, diluted 1:2, pH 4.5, information on percent solids is not available (the estimated concentration is around 20%)
444
JOSEPH SHILOACH AND ROBERT M. KENNEDY
Flow rate: 100-250 cm / h r Wash: 50 m M acetate buffer pH 4.5, containing 50 m M NaCl Elation: 100 m M phosphate buffer pH 9.0, containing 300 m M NaCl downward at 100 c m / h r d. Recombinant Human Nerve Growth Factor (rhNGF) from Cells^^
CHO
Absorbent: STREAMLINE SP in 25 m M MES/NaMES, 0.3 M sodium acetate pH 6.0 Capture conditions: temp. 37°C. Flow rate: 375 c m / h r Wash: 25 m M MES-NaMES, 0.3 M sodium acetate pH 6.0 Elution: 25 m M MES-NaMES 1 M sodium acetate pH 6.0. e. Recombinant Monoclonal Antibody from CHO Cell Culture^^
Adsorbent: STREAMLINE SP, 25 m M MES buffer pH 5.4 Capture conditions: the cell culture is kept for 2 to 3 days to settle the cells, then the supernatant, containing 5 X 10"^ cells/mL, is diluted 3 times and the pH adjusted to 5.4 Flow rate: 135-144 c m / h r Wash: 24 volumes of 25 m M MES buffer pH 5.4 Elution: linear gradient from 40 to 400 m M NaCl in 25 m M MES, pH 5.4 f. Humanized lgG4 Monoclonal Antibody from Myeloma Cell Cu/ture'^
Adsorbent: STREAMLINE Protein A in 50 m M glycine glycinate pH 8.0, containing 250 m M NaCl Capture conditions: the cell culture suspension is applied directly at 37°C. Flow rate: 300 c m / h r Wash: 50 m M glycine glycinate pH 8.0, containing 250 m M NaCl, and additional 10 column volumes, after the OD comes back to baseline Elution: 0.1 M glycine pH 3.0. iii. Discussion
As indicated in the previous section, the expanded-bed capture process for extracellular proteins from various sources is quite straightforward. Extracellular proteins from bacteria, mammalian cells and yeast can be captured, using different types of adsorbent materials, by directly pumping the cell suspension on the expanded-bed column. Preparations prior to the adsorbing process are minimal. For example, in the case of the mutant diphtheria toxin, culture dilution was needed; in the case of recombinant human serum albumin from P. pastor is ^ the culture was heat-treated; and in the case of monoclonal antibody from CHO cell, the culture was allowed to settle for 2 - 3 days before pumping it on to the anion-exchanger. Compared with the traditional approach (Fig. 1), the expanded-bed method is shorter and requires fewer steps (Table 4). The elimination of centrifugation, filtration, and ultrafiltration is likely to reduce processing time and capital ex-
EXPANDED-BED ADSORPTION
1 ^ 1
445
T A B L E 5 Economic Considerations for Obtaining C R M 9 Diphtheria Toxin"
Equipment Consumables Time (h)
Traditional
Expanded-bed adsorption
$90,000-$120,000^ $3,000-$4,300 (filters) 15 (set up and cleaning 10 hr, operation 5 hr)
$20,000'' $3,600 (resin) 14 (set up and cleaning 6 hr. operation 5 - 7 hr)
The comparison is for processing 70 liters of bacterial culture. Equipment needed: continuous centrifuge and tangential flow filtration system. Equipment needed: expanded bed column for 4.5 liter resin.
penses. Table 5 summarizes the economical considerations of the two approaches when mutant diphtheria toxin is recovered. B. Capture of Intracellular Proteins i. General Considerations
This category includes proteins that are produced inside microorganisms or mammaUan cells. To release these proteins, the biomass is first concentrated by conventional continuous centrifugation, and then it is disrupted. Therefore, the starting protein solution contains the targeted protein, other proteins, cell contents such as cell wall fragments and DNA, but it does not contain any media components. Unlike the solution that contains extracellular proteins, this solution is not homogenous, possibly affecting the bed stability; therefore, it is more difficult to achieve a stable expanded bed when dealing with intracellular proteins than when dealing with extracellular proteins. As previously stated, the bed stability is affected by the solution's ionic strength, pH, solids concentration and viscosity, and, in this case, also by the method of cell disruption. The disrupted cell suspension, after proper adjustment, is pumped directly on the expanded bed. The solid concentration in the suspension should be between 50 and 80 g per liter, the viscosity below 10 mPa, and the loading linear flow rate not less than 300 cm/hr. Unlike the extracellular protein case that deals with intact cells, in the disrupted cells case, the ionic strength has an effect on column stability. It is important to mix the loading solution continuously throughout the loading process to ensure homogeneous loading, and to adjust the binding conditions based on the properties of the targeted protein and the choice of matrix. ii. Practical Examples
This segment includes a detailed example of capture and recovery of intracellular protein (recombinant exotoxin A from E. coli) and key parameters and conditions for obtaining other products. a. Recombinant Pseudomonas aeruginosa Exotoxin A from £. co/i'^' '^
The modified P. aeruginosa exotoxin A is overproduced in £. coli BL21 (ADE3). The modified toxin, which lacks the enzymatic activity but retains the binding activity, is missing glutamic acid in position 553. The expression
446
JOSEPH SHILOACH AND ROBERT M. KENNEDY
4.5 kg Bacteria Sucrose20 mM Buffer90 I Suspension Centrifugation
2-3 h
Clarification
2-311
DEAE Sepharose fast flow 25.5 1 30 cm X 36 cm
loading 1h elution 2h .(100 cm h""")
I 36 1 F I G U R E 9 Flow diagram of the traditional capture process of recombinant exotoxin A from £ coli. (Reprinted from Johansson et o/.'^ with permission from Elsevier Science.)
of the modified toxin is controlled by the T7 promoter and the protein accumulates in the periplasmic space following induction with IPTG. When applying the traditional recovery process (Fig. 9), at the end of the fermentation, the cells are collected and suspended in two volumes of 20% sucrose in 20 m M Tris buffer pH 7.4 containing 1 m M EDTA. After the cells are well dispersed, they are mixed for an additional 10 min and then diluted with 18 volumes of 20 m M Tris buffer pH 7.4. The cell suspension is centrifuged and microfiltered through a 0.45 jubm filter. The clarified supernatant is applied on DEAE Sepharose Fast Flow at a flow rate of 100 c m / h r and a ratio of 40 OD280 nm per 1 mL. column. The elution is performed with a linear gradient of eight column volumes from 0 to 0.5 M NaCl in the Tris buffer. In contrast, when applying the expanded-bed process (Fig. 10), the collected cells are suspended in 2 volumes of 20% sucrose in 50 m M Tris buffer pH 7.4 containing 1 m M EDTA. After the cells are well dispersed.
4.5 kg E Bacteria Sucrose-
•
50 mM buffer180ISu spension
.
.
(400 cm h ') 2 1 min ^ loading-1801 washing - 40 1 2h
i Streamline DEAE 4.71 20 cm X 15 cm
• 0.5 NaCI 0.5 I min""" (100 cm h-'')
t ,I \
13 51 F I G U R E 10 Flow diagram of the expanded-bed capture process of recombinant exotoxin A from £. coli. (Reprinted from Johansson et 0/.'^ with permission from Elsevier Science.)
EXPANDED-BED ADSORPTION m
447
T A B L E 6 Capture Step Comparison for Processing 4.5 Kg £. Co/f Cells for Production of Exotoxin A 553D a
Packed-bed DEAE Sepharose FF Specific activity after the step (mg toxin/mg protein) Recovery (%) Processing time (hr) Column volume (liters) Eluent volume (liters)
0.1 73.0 8.0 25.5 36.0
Expanded-bed STREAMLINE DEAE 0.06 79.0 2.5 4.7 13.5
Reprinted from Johansson et al}^ W\t\v permission from Elsevier Science. Packed column numbers are extrapolated values.
they are mixed for an additional 10 min and then are diluted with 18 volumes of 50 m M Tris buffer, pH 7.4. The mixture is mixed for an additional 10 min, endonuclease (Benzonase) is added at a ratio of 7S units per gram cells, and the suspension is diluted with 18 volumes of the Tris buffer. The cell suspension is then pumped upward on an equilibrated and expanded STREAMLINE DEAE at a ratio of 1 g cell extract per 1 mL packed column at a flow rate of 400 cm/hr. After all the cell suspension is applied on the column, the column is washed with 50 m M Tris buffer pH 7.4 until the OD at 280 m M of the eluent is back to baseline. At this point, the upward flow is stopped, the bed is allowed to settle, and the flow adaptor is lowered to the surface of the packed bed. After washing in a downward mode with 2 column volumes at a rate of 100 cm/hr, the protein is eluted with 20 m M Tris buffer pH 7.4 containing 0.5 M NaCl. The OD280 P^ak is collected for the next purification step. Successful adsorption procedure is achieved using the following conditions: concentration of the dry cell mass is 6 g/liter, the buffer concentration is 50 m M and the endonuclease content is 75 units per gram cells. The comparison between the traditional process and the expended-bed process is summarized in Table 6. b. Recombinant
Diphtheria
Toxin from £. c o / i ' *
Method of cell disruption: £. colt cells are suspended in the binding buffer (0.25 gram cells per mL), and the suspension is passed twice through a homogenizer (Manton Gaulin) at 9000 psi. Absorbent: STREAMLINE Chelating, saturated with NiS04 and equiUbrated in 20 m M Tris, pH 8.0 containing 0.5 M NaCl and 5 m M imidazole. (The protein is expressed with His-tag). Capturing conditions: broken cell suspension at a concentration of 0.13 g / m L in binding buffer. Flow rate: 200 m L / h Wash: 20 m M Tris pH 7.2 containing 0.5 M NaCl and 30 m M imidazole.
448
JOSEPH SHILOACH AND ROBERT M. KENNEDY
Elution: 20 m M Tris pH 7.2, containing 0.5 M NaCl and 250 m M imidazole c. Recombinant Annexin V from £. coli^^
Method of cell disruption: cell suspension is passed three times through a high-pressure homogenizer 10,500-13,500 psig. Adsorbent: STREAMLINE DEAE equilibrated in 30 m M ammonium acetate pH 5.5 Capturing conditions: broken cell suspension at a concentration of 3.6% solids in 30 m M ammonium acetate pH 5.5; the suspension contains 1% detergent Triton X-100 Flow rate: 300 c m / h r Wash: 30 m M ammonium acetate pH 5.5 Elution: 30 m M ammonium acetate pH 5.5 containing 250 m M NaCl 100 c m / h r d. Recombinant Anti-HIV Fab-Fragment from £. co/i'®
Method of cell disruption: cell suspension passed three times through high-pressure homogenizer at 10,000 psig Adsorbent: STREAMLINE SP in 50 m M sodium acetate pH 5.0 Capturing conditions: dry cell concentration, 1.4% suspended in 50 m M sodium acetate pH 5.0, containing endonuclease (benzonase) at a ratio of 10 fiL per 60 g cells (dry weight). Flow rate: 300 c m / h r Washing: 50 m M sodium acetate pH 5.0 Elution: 50 m M sodium acetate pH 5.0, containing 1 M NaCl, at 100 cm/hr e. Recombinant lnterluldn-8 from £. co/i Inclusion Bodies
Method of cell disruption: cell pellet suspended in three volumes of 30 m M sodium phosphate pH 6.5 containing 6 M guanidine hydrochloride. Suspension is mixed for 3 hr and diluted with six volumes of water in two steps: three volumes in the first step followed by mixing for 30 min, and three volumes in the second step followed by mixing overnight. Adsorbent: STREAMLINE SP in 30 m M sodium phosphate pH 6.5 Capturing conditions: 1% (dry weight) cell suspension in 30 m M sodium phosphate containing 6 M guanidine hydrochloride Flow rate: 300 c m / h r Washing: 30 m M sodium phosphate containing 6 M guanidine hydrochloride Elution: 30 m M sodium phosphate containing 6 M guanidine hydrochloride, and 0.5 M NaCl at a flow rate of 100 c m / h r f. Glucose 6 Phosphate Dehydrogenase (G6PDH) from S. cerevisiae^^
Method of cell disruption: yeast suspension 50% wet weight is passed through a bead mill using 0.5 mm glass beads
449
EXPANDED-BED ADSORPTION
Adsorbent: STREAMLINE DEAE in 50 m M sodium phosphate pH 6.0 Capturing conditions: 6.5% (dry weight) yeast suspension in 50 m M sodium phosphate pH 6.0, viscosity 5.0 mPa Flow rate: 200 c m / h r Washing: 2 5 % glycerol in 50 m M sodium phosphate pH 6.0 in an expanded mode, followed by 50 m M NaCl in 50 m M sodium phosphate pH 6.0 in a packed mode Elution: 150 m M NaCl in 50 m M sodium phosphate pH 6.0 iii. Discussion
Capturing intracellular proteins on an expanded bed involves preliminary steps that are not required for the capture of extracellular proteins. These steps are the initial cell concentration, usually by filtration or centrifugation, and cell disruption, using a high-pressure homogenizer, a bead mill, or osmotic shock. To prepare a disrupted cell suspension for direct adsorption on an expanded bed, suspension viscosity and ionic strength can be adjusted by adding detergent, endonuclease, and salt. It is also important to adjust the concentration of solids in the broken cell suspension because it affects the stability of the expanded matrix. As seen in the example of capturing recombinant Interlukin-8 from £. coli, the expanded-bed process is also suitable for capturing a protein extracted from inclusion bodies using guanidine hydrochloride. Since concentration of the cells is an essential preliminary step when processing intracellular proteins, and instrumentation for this unit operation is required, a comparson of the economical considerations between the described approach and the traditional method is different from the case of the extracellular proteins (Table 7).
Yll. DISCUSSION AND CONCLUSIONS The production examples described here, and the information published since this chapter was written, clearly indicate that the expanded-bed operation offers an efficient alternative to the conventional protein capturing process.
T A B L E 7 Economic Considerations for Obtaining Recombinant Exotoxin A (533D)°
Equipment Consumables Time (hr)
Traditional
Expanded-bed adsorption
$103,000 $1,500 (filters) $10,000 (resin) 20 (set up and cleaning 11 hr. operation 8-9 hr)
$90,000"" $3,600 (resin) 12.5 (set up and cleaning 6.5 hr. operation 6 hr)
The comparison is for processing 4.5 kg of bacterial biomass. Equipment needed: continuous centrifuge and chromatography column. Equipment needed: continuous centrifuged and expanded-bed column for 4.5 liter
450
JOSEPH SHILOACH AND ROBERT M. KENNEDY
The latter requires that the starting product solution be centrifuged, clarified, and at times, dialyzed, before it can be adsorbed on a packed column. In contrast, capturing the desired product by the expanded-bed method is simpler because the crude starting solution is pumped directly on the adsorbent. To achieve a successful expanded-bed process operation, the adsorbent must be stable in its expanded mode, requiring that careful evaluation of the loading conditions take place during the process development stage. The follow^ing points should be considered: establishing optimal loading conditions for intracellular proteins was found to be more complicated than for extracellular proteins; starting solutions with high viscosity and solid content often must be diluted; and when the loading solution is very crude, the upper flow adaptor may get clogged, requiring that the flow direction be changed frequently. Note that the expanded-bed approach is not a viable option if the binding of the desired product to the adsorbent is possible only after initial partial purification. In summary, when evaluating the suitability of the expanded-bed approach for capturing a particular protein, one must consider whether the product is from bacteria, yeast, or mammalian cells; whether it is intracellular or extracellular; and the specific activity of the final product, its overall yield, the processing time and capital, labor, and maintenance costs.
REFERENCES 1. Chase, H. A. (1994). Purification of proteins by adsorption chromatography in expanded beds. Trends Biotechnol. 12, 296-303. 2. Hjorth, R. (1997). Expanded bed adsorption in industrial bioprocessing: Recent developments. Trends Biotechnol. 15, 230-235. 3. Gailliot, E. P., Gleason, C , Wilson, J., and Zwarick, J. (1990). Fluidized bed adsorption for whole broth extraction. Biotechnol. Prog. 6, 370-375. 4. Thommes, J. (1997). Fluidized bed adsorption as a primary recovery step in protein purification. Adv. Biochem. Eng. Biotechnol. 58, 185-230. 5. Amersham Pharmacia (1995). "Application Note: Cleaning in Place," Publ. No. 18-1115-27. 6. Hanson, K. A. (1995). Physical chemical properties of STREAMLINE Ion exchangers. Poster Presentation, Eur. Congr. Biotechnol., 7th, Nice, France. 7. Fass, R., Bahar, S., Kaufman, J., and Shiloach, J. (1995). High yield production of diphtheria toxin mutants by high density culture of C7(6)*'^^^ strains grov^n in a non deferrated-media. Appl. Microbiol. Biotechnol. 43, 83-88. 8. Shiloach, J., and Kaufman, J. B. (1999). The combined use of expanded bed adsorption and gradient elution for capture and partial purification of mutant diphtheria toxin (CRM 9) from Cory neb acterium diphtheriae. Sep. Sci. Technol. 34, 29-40. 9. Zurek, C , Kubis, E., Keup, P., Hoerlein, D., Beunink, J., Thoemmes, J., Kula, M. R., Hollenberg, C. P., and Gellissen, G. (1996). Production of two Aprotonin variants in Hansenula plymorpha. Process Biochem. 31, 679-689. 10. Noda, M., Sumi, A., Ohumura, T., and Yokoyama, K. (1966). Process for purifying recombinant human serum albumin. Eur. Pat. Appl. EP 0 699 687 A2. 11. Beck, J., Liten, A., Viswanathan, S., Emery, C , and Builder, S. (1996). Direct capture of Nerve Grov^th Factor from CHO cell culture by EBA. Presentation, Recovery Biol. Prod., 8th, Tucson, AZ. 12. Batt, B. v., Yabannavar, V. M., and Singh V. (1995). Expanded bed adsorption process for protein recovery from whole mammalian cell culture broth. Bioseparation 5, 4 2 - 5 3 .
EXPANDED-BED ADSORPTION
45 I
13. Jagerstern, C , Johansson, S., Bonnerjea, J., and Pardon, R. (1996), Capture of a humanized IgG4 directly from the fermentor using STREAMLINE r-protein A. Presentation, Recovery of Biol. Prod., 8th, Tucson, AZ. 14. Fass, R., Van de Walk, M., Shiloach, A., Joslyn, A., Kaufman, J., and Shiloach, J. (1991). Use of high-density cultures of Escherichia coli for high level production of recombinant Pseudomonas aeruginosa exotoxin A. Appl. Microbiol. Biotechnol. 36, 65-69. 15. Johansson, H. J., Jagersten, C , and Shiloach, J. (1996). Large-scale recovery and purification of periplasmic recombinant protein from E. coli using expanded bed adsorption chromatography follow^ed by new ion exchange media. / . Biotechnol. 48, 9-14. 16. Noronha, S., Kaufman, J., and Shiloach, J. (1999). Use of STREAMLINE Chelating for capture and purification of poly His Tagged recombinant proteins. Bioseparation 8, 145-151. 17. Barnfield Frej, A. K., Hjorth, R., and Hammarstsrom A. (1994). Pilot scale recovery of recombinant Annexin V from unclarified £. coli homogenate using expanded bed adsorption. Biotechnol. Bioeng. 44, 922-929. 18. Jagerstern, C. (1994). Purification of recombinant Anti-HIV Fab-fragment expressed in £. coli. Presentation, Recovery Biol. Prod., 7th San Diego, CA. 19. Chang, Y. K., and Chase, H. A. (1966). Ion-exchange purification of G6PDH from unclarified yeast cell homogenate using expanded bed adsorption. Biotechnol. Bioeng. 49, 204-216.