Journal of Non-Equilibrium Thermodynamics2007-32!99!127

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

J. Non-Equilib. Thermodyn.

2007  Vol. 32  pp. 99–127

Review Article

Isolation and Purification of Biotechnological


Products
Jürgen Hubbuch1,* and Maria-Regina Kula2
1 Institut für Biotechnologie 2, Forschungszentrum Jülich, 52425 Jülich, Germany
2 Institut für Enzymtechnologie, Heinrich-Heine-Universität Düsseldorf, 52426

Jülich, Germany

*Corresponding author (j.hubbuch@fz-juelich.de)

Abstract
The production of modern pharma proteins is one of the most rapid growing
fields in biotechnology. The overall development and production is a complex
task ranging from strain development and cultivation to the purification and
formulation of the drug. Downstream processing, however, still accounts for
the major part of production costs. This is mainly due to the high demands
on purity and thus safety of the final product and results in processes with
a sequence of typically more than 10 unit operations. Consequently, even if
each process step would operate at near optimal yield, a very significant
amount of product would be lost. The majority of unit operations applied
in downstream processing have a long history in the field of chemical and
process engineering; nevertheless, mathematical descriptions of the respective
processes and the economical large-scale production of modern pharmaceuti-
cal products are hampered by the complexity of the biological feedstock, es-
pecially the high molecular weight and limited stability of proteins. In order
to develop new operational steps as well as a successful overall process, it is
thus a necessary prerequisite to develop a deeper understanding of the ther-
modynamics and physics behind the applied processes as well as the implica-
tions for the product.

1. Introduction
Mankind harnessed biological production first in the form of agriculture and
animal husbandry und used it for thousands of years to sustain food supplies,
which was the basis for settlement and culture. Empirical use of biocatalysts

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


6 Copyright 2007 Walter de Gruyter  Berlin  New York. DOI 10.1515/JNETDY.2007.004
100 J. Hubbuch and M.-R. Kula

in food preservation and food production from agricultural raw material also
dates back a few thousand years, e.g., in cheese manufacturing, brewing and
baking, as well as lactic acid fermentation of milk and vegetables. In the be-
ginning of the last century, microbial production of citric acid and enzymes
by surface culture were the first industrial biotechnological products outside
food and feed production after biologists learned to isolate and maintain
pure microbial cultures. The discovery and production of antibiotics was the
next major breakthrough, because during these developments the aseptic sub-
merse cultivation in large bioreactors was established. The new technology
was quickly expanded to produce bulk quantities of extracellular enzymes by
selected strains of Bacillus, Aspergillus and other microorganisms.
In parallel, early in the twentieth century vaccine production was developed;
the first proteo-hormon, insulin, was isolated from pig pancreas for the treat-
ment of diabetes patients; and human albumin was isolated from donated
blood in large scale for use as plasma expander. At the time, these ‘‘biologi-
cal’’ drugs could not be analyzed by conventional means to ensure their
identity and safety. Therefore, a framework was developed that defined a bi-
ological drug by the method of its production. All this actually happened be-
fore even the chemical nature of proteins was finally proven by Sanger and
Thompson in 1953 [1].
The next large impact on biotechnological production arose from the advances
in gene technology. Given the largely universal nature of the genetic code, in
principle it is possible today to transfer a piece of DNA containing the infor-
mation for a certain protein into a di¤erent organism and use the machinery
of the host cell to synthesize that protein. The diabetes drug human insulin,
for example, is produced now in the bacterium Escherichia coli or in the yeast
Saccharomyces cerevisiae. Together with the advancing insights gained into
the cause of disease, we find an increasing number of pharma proteins on
the market and much more under development for various applications. Not
only has the pharmaceutical sector profited from gene technology, but en-
zyme production, biocatalysis, and biotransformations have changed dramat-
ically. Another important field of application is metabolic engineering, when
throughput through existing metabolic pathways is intensified for the produc-
tion of small molecules such as hydroxy acids or amino acids or even new
pathways constructed to produce diols or indigo in bacteria. Biological pro-
duction utilizes mainly sugar as raw material and therefore is a sustainable
technology based on renewable resources, which should help to gradually re-
place petrochemicals as raw materials in the not so distant future.
Biological synthesis of goods, however, is not the only part of a production
process. Before it can be o¤ered on the market, invariably the product of in-
terest has to be separated from the biomass and more or less purified –

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 101

Table 1 Range of products produced in the biotech industry di¤ering widely in size, chemical
nature, and stability.
Type of product Examples
Whole cells Baker’s yeast, other starter cultures
Primary metabolites Ethanol, organic acids, amino acids
Secondary metabolites Antibiotics, steroids, pharmaceutical intermediates
Carbohydrates Xantan, dextran
Proteins Bulk enzymes, specialty enzymes, proteo-hormons, monoclonal
antibodies, growth factors, virus-like particles
Nucleic acids Plasmid DNA, RNA

depending on its application. With the increasing e‰ciency and flexibility of


biological synthesis, the contribution of the purification cost to the final price
of goods is rising steeply. Ethanol synthesis from sugar is highly e‰cient us-
ing yeast or bacteria, but isolation from a dilute and aqueous feedstream is
expensive; in consequence, only alcoholic beverages selling water together
with ethanol was profitable on a large scale for a long time. In recent years,
more ethanol produced from starch is entering the market as a gasoline addi-
tive and fuel. In the excitement over the tremendous progress in biological
synthesis, the downstream processing of products has been sadly neglected.
To realize the full potential of biotechnological production, separation and
purification issues have to be resolved. Table 1 illustrates the kind of product
encountered; it is obvious that there is no single solution for recovery and
purification. The nature of products and their physical properties are too di-
versified. Focusing on high molecular weight products, Figure 1 provides a
general flow sheet depicting di¤erent stages in downstream processing and
identifies unit operations that may be employed for the respective tasks. De-
pending on the biological system and product, a rough distinction can be
made between the recovery of whole cells, e.g., use as single cell protein,
where simple solid–liquid separation leads to the desired product and the
more advanced systems requiring a complex process train in order to reach
the desired purity. Prior to the more advanced sections starting with the iso-
lation of the product from the initial clarified process fluid, di¤erent strategies
are necessary to prepare the cultivation broth. Biological expression systems
secreting the product through the membranes into the culture medium require
a simple solid–liquid separation, while for proteins expressed intracellularly,
a cell homogenization step has to be included in order to liberate the product.
Expression as an inclusion body increases the complexity by one step toward
a solid–solid–liquid separation, where cell debris, inclusion bodies, and pro-
cess liquid have to be separated. Following this, the di¤erent stages depicted
in Figure 1 represent the tasks of product isolation from the processing fluid
and consequent concentration, the purification from soluble contaminants, as
well as polishing to meet purity demands, storage requirements, and product

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


102 J. Hubbuch and M.-R. Kula

Figure 1 Schematic overview of di¤erent stages during downstream processing (shaded gray)
and the respective unit operations commonly applied during these stages. Initial processing is
highly dependent on the product itself and how it is expressed (light gray). ATPS, aqueous two
phase system; EBA, expanded bed adsorption; HGMS, high gradient magnetic separation.

formulation. Along this process train, certain unit operations are typically
found to meet the resolution needed. A selection of the most prominent unit
operations and their placement within the process is given at the top and bot-
tom of the respective sections.

Process design still relies to a considerable extent on experience and heuristic


rules [2]. In order to reduce the cost, it is important to lower the number of
steps needed to achieve the desired purity of the product as well as to improve
the e‰ciency and yield of each single step. Besides these economic considera-
tions, our understanding and consequently the modeling of the di¤erent unit
operations and the complex interactions between various parameters need to
be considered and advanced. There are two areas where progress is especially
important for future developments. The production of small molecules as
commodities is characterized by a high contribution of the raw material cost
to the final cost of goods, and post-synthesis processing has to be highly e‰-
cient and cheap to compete with chemical synthesis, as illustrated by ethanol
production. The rising cost of petrochemicals reduces the economic pressure;
nevertheless, improvements are mandatory for the success of many new bio-
technological production routes for bulk products. At the other extreme
we find antibody-based pharmaceuticals, where highly purified proteins
are administered in rather high dosages for long times, leading to production

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 103

predictions up to 1000 to/a for single products. In this case, raw material
costs are comparatively low, even if these glycosylated antibodies are pro-
duced in animal cell cultures, but yield and purification costs are significant.
For the majority of patients, drug prices are regulated by health insurance or
state authorities and in order to reach these patients the final price has to be
acceptable, putting heavy demands on the technical as well as economical
performance of purification operations. In the following sections the di¤erent
stages in downstream processing of biotechnological products will be briefly
discussed.

2. Protein stability
With the onset of recombinant protein production in micro-organisms
and their subsequent release and purification, protein stability during down-
stream processing has become one of the major issues determining the success
of a given process. The potential factors for product degradation can be di-
vided into physical processes (shear, impingement, aggregation, precipita-
tion), fluid phase system parameters (pH, ionic strength, bu¤er composition,
temperature), biological (enzymatic degradation) or chemical processes (oxi-
dation, deamidation, hydrolysis) [3]. With the high number of unit operations
in a conventional purification process, holding times, and changes in pH,
temperature, and solvent composition, it becomes clear that detailed knowl-
edge on protein stability on a molecular basis is needed. This molecular
understanding of structural changes of biomolecules as a result of environ-
mental influences, however, still represents a white area in today’s process
design.

3. Cell disintegration
Depending on the location of the respective product, a release of the product
from the cell might be a necessary first step in a purification process. Micro-
bial cell walls are composite materials of carbohydrates and peptide or pro-
teins exhibiting considerable mechanical strength comparable to reinforced
concrete. To release soluble products as well as inclusion bodies, the cell wall
has to be broken first. A good review on various techniques for cell disinte-
gration is given by Middelberg [4]. While a number of potential process tech-
niques exist at small scale – mainly used in microbiological laboratory appli-
cations – at large scale only few approaches are feasible (see Figure 2) [5].
Generally, one distinguishes between processes where mechanical shear is
applied for the breakage of the cell wall and more refined processes where
specific targets on the cell wall are attacked (lysis) in order to weaken and

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


104 J. Hubbuch and M.-R. Kula

Figure 2 Schematic overview of various methods for cell disruption. Underlined methods are
currently used on a large scale in industrial production [5]. *Main application of alkaline lysis:
pDNA isolation; enzymatic lysis using lysozyme: glucose isomerase purification.

consequently open the cell wall. The most common unit operations are high-
pressure homogenization or wet milling. High-pressure homogenizers are orig-
inally adapted from the milk industry or wet mills originally developed for o¤-
set dyes. In both types of equipment, heat generation and heat removal have to
be taken into account. With the onset of gene therapy trials, the chemical pro-
cess consisting of an alkaline lysis of E. coli cells containing pDNA has gained
industrial importance. This is mainly due to its excellent solubilization of the
cell wall, quantitative release of intracellular products combined with a selec-
tive renaturation of pDNA, while most contaminants remain in a precipi-
tated form after neutralization [6–9]. This process serves also as a good ex-
ample for a system where classical approaches based on mechanical cell
disruption would lead to shear-induced degradation of the product [10, 11].

The success of the unit operations following cell disintegration steps is closely
linked to the homogenization performance, and it is evident that the desired
properties of the cell debris are di¤erent when considering filtration, centrifu-
gation, or expanded bed adsorption as the unit operation of choice. It has
been recently shown that, besides a quantitative release of the target product,
parameters such as debris size and charge as well as fluid phase parameters
determine to a high degree potential fouling processes and thus the ultimate
performance of initial steps [12–14].

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 105

4. Solid–liquid separation

4.1. General aspects


During biosynthesis, products are either accumulated in soluble or insoluble
form within the cells or selectively secreted into the culture medium. In case
of a selective secretion, the solid–liquid separation aims at a quantitative re-
covery of the fluid phase. If products are accumulated in the cell, a certain
concentration and separation from the cultivation broth can be achieved by
an initial cell separation step. However, after product release by cell disrup-
tion, a second, generally more di‰cult solid–liquid separation step for re-
moval of the cell debris from the product of interest is created.
The rather small size of bacteria (1–3 mm), yeast (4–8 mm), and mammalian
cells (ca 20 mm), their low density di¤erence to aqueous media, as well as their
easy deformation and high water-binding capacity, make the removal of cells
or cell debris from culture fluid quite di‰cult on a large scale using conven-
tional unit operations such as centrifugation or filtration. Furthermore, these
classical unit operations solely aim at solid–liquid separation and thus do not
allow selective concentration or purification of the released product. These
applications arise mainly from the fields of mechanical process engineering
and fluid dynamics and are, due to the complexity of the fluid, by a first
approach described by basic dependencies such as the relationships describing
the settling velocity of a solid in case of sedimentation processes (Eqs. 1–3) or
the Darcy relationship describing the performance of filtration steps (Eq. 4).

In order to shorten process trains by fusing two or more formerly separated


tasks, developments of integrated capture steps combining solid–liquid sepa-
ration with a thermodynamically controlled step for product capture and iso-
lation have received much attention in recent years. Figure 3 provides an
overview of di¤erent processes used for solid–liquid separation in the biotech
industry and how they relate to each other in terms of their ability to cope
with the biomass load applied and their potential for further concentration
of the solids [15].
Finally, solid–liquid separation is also an integral part of crystallization or
precipitation operations for the collection of crystals or precipitates.

4.2. Centrifugation
Centrifugation is a well-established technology widely used in various
industries. The method employs the di¤erence in sedimentation velocities
of particles in a given system. In the creeping flow range (Re a 0:2) with
no external influencing parameters other than fluid viscosity, density, and

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


106 J. Hubbuch and M.-R. Kula

Figure 3 Overview of di¤erent methods for solid–liquid separation in downstream processing


and their ability to cope with solid concentration as well as to further concentrate incoming
solids (reproduced with kind permission from [15]). EBA, expanded bed adsorption; HyCon,
hyper-concentrator centrifuge (Westfalia, Oelde, Germany).

gravitational acceleration, Stokes law presents a simple equation describing


such an idealized settling velocity:

ds2 ðrs  rf Þ
vsed ¼ g ð1Þ
18 h

In a technological system, the above constraints are naturally not given and
influences from other particles present, their size heterogeneity, as well as fric-
tion forces from boundaries in the machinery will dominate the performance.
ro 2
Nevertheless, by multiplying Eq. 1 by the acceleration factor v ¼ describ-
g
ing the ratio of centrifugal acceleration ro 2 to gravitational acceleration g, we
obtain a simple equation describing the basic parameter dependencies in a
laminar flow field:

ds2 ðrs  rf Þ 2
vsed ¼ ro ; ð2Þ
18 vh

where vsed is the speed of sedimentation, ds the particle diameter, rs and rf


the density of the solids and the liquid phase, r the radius of rotation, o the
angular velocity, and h the dynamic viscosity. The density di¤erence provides

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 107

the driving force of the separation. Because of the high water content of cells,
the density di¤erence is rather small in aqueous media. The higher density
of protein aggregates is exploited in the centrifugal separation of so-called
‘‘inclusion bodies’’ from cell debris when the product is obtained as inclusion
bodies, e.g., in recombinant E. coli cultures [16]. The dependency of the
throughput on feed properties and machine as well as operating parameters
is given by Eq. 3.

Q ¼ vsed; g S S ¼ vA; ð3Þ

with Q as the volumetric feed flow, vsed; g the Stokes settling velocity in a grav-
itational field. The sigma factor S is also called the ‘‘equivalent clarification
area’’ describing the settling area A of a gravity-settling tank reaching the
same clarification capacity as the centrifuge in question. For a given through-
put, the S factor methodology o¤ers a first rule of thumb when performing
scale changes for a given process.
In order to increase productivity and product titers, recent improvements in
cultivation technology resulted in high cell-density cultures containing up to
60% wet weight cells (Figure 3) but exhibiting much higher viscosities in the
broth. According to Eq. (2), higher viscosities reduce the performance of cen-
trifugation steps and for secreted products lead to increased losses in the in-
terstitial space unless a cumbersome washing step is introduced.
For the processing of large volumes, continuously operating disk stack centri-
fuges with intermittent solid discharge or nozzle design are mainly employed
in the biotech industry. Disk stack separators are run at g-forces up to 15 000
g; bowls with up to 60 cm diameter are available. The further scale up of
these machines is limited by properties of the construction material of the ro-
tor bowl as well as demands on energy transfer during operation. Depending
on the volume and properties of the feed, often it may be necessary to install
multiple centrifuges for parallel processing to meet time constraints. The
maximal volumetric capacity for E. coli broth is approximately 2,500 l per
hour for moderate viscosity and will be lower for higher cell densities.
The shear sensitivity of mammalian cells led to the development of specially
designed acceleration zones in disk stack centrifuges in order to avoid exten-
sive cell damage. In addition to standard industrial centrifuges, equipment for
the biotech industry is designed for cleaning in place, sterilization, and con-
tainment if needed. An extensive review on di¤erent solid–liquid separation
techniques for mammalian cell cultures can be found by Voisard et al. [17].
Decanters are today mainly employed in biological sewage plants. Solid re-
moval in a decanter is less e¤ective; therefore, a pretreatment of the feed is
usually needed to induce flocculation and increase the diameter of flocks. The

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


108 J. Hubbuch and M.-R. Kula

auxiliary chemicals introduced for that purpose, e.g., multivalent salts or poly-
electrolytes, should not interfere with product purification, which limits this
approach considerably.

4.3. Filtration
Filtration is a pressure-driven process and is used in many di¤erent places in
the isolation of biological products for solid–liquid separation. The first de-
scription was given by Darcy already in 1856, relating various parameters of
a cake filtration encountered in water purification [18]:

Dp h dV
¼ ; ð4Þ
l kA dt

where Dp describes the pressure drop over a length l for a fluid with a viscos-
ity h provided a filter with cross-sectional area A and permeability k at a vol-
umetric throughput dV =dt.
Owing to the compressibility and large water retention of cells, cake filtration
in flow-through mode is only used in special cases, e.g., harvesting mycelia
and for final clarification of process liquor, if the solid load is low. Besides,
filtration is employed to collect crystals such as amino acids. Today filtration
devices are often operated with tangential flow using convective transport
especially in protein purification, thereby avoiding cake formation as far as
possible and improving throughput [19, 20]. The principle and potential fac-
tors causing a decrease in filtration performance are illustrated in Figure 4.
Another approach to increase performance of a filtration process uses filter
aids to improve the mechanical stability of the cake formed and to allow op-
eration at higher pressure, which was widely employed in the biotech indus-
try, e.g., in the isolation of antibiotics or bulk enzymes. The cost of treatment
for spend filter aid and/or the demand for contained operation limits the ap-
plicability of this approach today, and in new productions alternative tech-
nology is usually chosen.

5. Integrated unit operations

5.1. General aspects


To overcome some of the inherent di‰culties in handling cells and cell debris,
new approaches try to combine a thermodynamically controlled process with
the solid–liquid separation step and achieve clarification, concentration, and
a moderate purification of the product in a single operation [21]. This can be
carried out either by extraction or by adsorption under selected conditions.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 109

Figure 4 Principle of cross-flow filtration and factors a¤ecting filtration performance [100].

5.2. Extraction
Extraction is widely used in the chemical industry; the theoretical understand-
ing is well advanced for small molecules extracted into organic solvents. Be-
sides, a wide selection of equipment for batch and continuous processing is
commercially available. In the biotechnology industry, extraction is used in
the isolation of antibiotics, e.g., penicillin from crude or pre-clarified broth.
Reactive extraction, which exploits the formation of ion pairs to transport
hydrophilic molecules such as organic acids into the organic phase, is a newer
development in order to reduce the environmental burden associated with
waste production in traditional processing [22].
Most proteins, however, are not soluble in organic solvents, at least not in
an active conformation. Nevertheless, protein extraction is possible making
use of aqueous two-phase systems [23, 24]. Two immiscible, aqueous phases
form by incompatibility if the concentration of two hydrophilic polymers
such as polyethylen glycol and dextran exceeds certain threshold values in
the common solvent water, or by salting out of polyethylen glycol from aque-
ous solutions selecting an appropriate salt and concentration. In thermosensi-
tive systems, based on surfactants or copolymers, phase separation can be
brought about by raising the temperature above the cloud point. The technol-
ogy uses available commercial equipment. The selection of a phase system for

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


110 J. Hubbuch and M.-R. Kula

extraction depends on properties of the target molecule [25–27]. Protein ex-


traction has been extensively studied including scale up to pilot and industrial
operation [28, 29]. Phase separation in aqueous two-phase systems is quite
well understood by now. The factors governing protein partition are complex
and numerous, however; therefore, only heuristic rules exist today for process
development, hampering application in industry. In view of the amounts of
proteins envisaged for the future, several aspects make protein extraction at-
tractive: the demonstrated possibility of continuous processing [30], the abil-
ity to handle high cell density cultures without prior dilution [31], the ease of
scale up and the gentle environment leading to high activity yields. Besides,
mathematical optimization methods and automated high-throughput screen-
ing devices have been introduced recently, which allow the evaluation of a
large parameter set in a short time and overcome the lack of understanding
by empirical optimization [32, 33].

5.3. Adsorption
Depending on reaction parameters, the interaction of a molecule with a
surface may lead to binding; the process is called adsorption. For separation
purposes, only reversible adsorption is of interest. The latter requirement ne-
cessitates su‰ciently hydrophilic surfaces handling proteins. In many cases,
reversible adsorption can be modeled using Eq. 5 derived by Langmuir as-
suming single layer formation, 1 : 1 stoichiometry, and non-overlapping bind-
ing sites.

c
q  ¼ qmax ð5Þ
kd þ c 

The isotherm given by Eq. 5 depends on solution parameters such as pH,


ionic strength, and selection of bu¤er ions, as well as properties of product
and surface. Many forces contribute to binding: electrostatic attraction and
dipol–dipol interaction, hydrophobicity, van der Waals forces, etc. By cova-
lent modification of the surface with appropriate ligands, adsorbents are pro-
duced in which one of those forces is expected to dominate. Adsorption is
governed by the equilibrium, e.g., as described by Eq. 5, but transport to the
ligand is very important for the kinetics of binding. Most preparative chro-
matographic adsorbents are based on porous materials providing large inter-
nal surfaces for binding but in consequence will exhibit pore di¤usion limita-
tions especially in the capture and purification of large molecules [34]. A
principle overview over commonly used base materials, adsorbent designs,
and interaction modes commonly used in bio-chromatography is given in
Table 2. In order to circumvent the limitations of di¤usive transport, new

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Table 2 Adsorbent types used for chromatographic protein separation. For a detailed description, see 5.3.1 expanded bed adsorption and 8.2
J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2

Isolation and Purification of Biotechnological Products


packed bed chromatography.
Type Backbone material Resin/pore design Interaction mode/ligands
Packed bed
Chromatography
(see 8.2.)

Agarose Open pore structure (size) Ion exchange


Cellulose Di¤usive pores Hydrophobic
Dextran Convective flow pores Mixed mode
Divinylbenzol Product-related pore sizes Thiophilic
Hydroxyapatite Pore design (ligand placement) Reversed phase
Methacrylamide Open pores A‰nity
Methacrylate Polymer chains
Silica Gel in a shell
Styrol
Expanded bed adsorption
(see 5.3.1)

(a) Ceramics (a, b) Open pore/composite structure Ion exchange


(b) Dextran/quartz core (c) Shell protection Hydrophobic
(b) Agarose/stainless steel (d) Surface grafting Mixed mode
Density range: Thiophilic
r: 1.2–3.2 g/ml A‰nity
Size range:
Daverage : 75–135 mm

111
112 J. Hubbuch and M.-R. Kula

approaches using either nonporous adsorbent materials or materials with


larger pores enabling convective transport of the molecule to the ligand are
pursued. The need to handle solids containing feedstocks when integrating
solid–liquid separation with an adsorptive-based primary recovery of the
product resulted in three principle modes of process operation: expanded
bed adsorption (EBA) as a fluidized bed of chromatographic adsorbents,
membrane adsorber with a large flow through pores, and high gradient mag-
netic separation (HGMS) where product capture and solid–liquid separation
has been uncoupled.

5.3.1. Expanded bed adsorption The combination of classical chromatogra-


phy with solid–liquid separation using a stabilized fluidized bed of chromato-
graphic adsorbents is termed expanded bed adsorption [35]. The fluidization
leads to an increase of inter-particle voidage – as compared to a packed bed –
and thus allows solids to pass through the column while at the same time act-
ing as a chromatographic column. In order to combine relatively high flow
rates (3 m/h P 12 m/h) and viscosities, the respective resins have to express a
relatively high density when compared to packed bed adsorbents. This high
adsorbent density has to be ensured using either high-density material or in-
corporating a core material (e.g., quartz, steel, etc.) into the respective resin
(see Table 2). Therefore, adsorption of product is possible from unclarified
feed streams – either cell culture or cell homogenates – under certain bound-
ary conditions. It has, however, been shown during the development of EBA
that a successful application of this technique was strongly linked to the abil-
ity of handling cells and cell debris during processing. One approach to the
bead architecture (shell protection) was the introduction of a protecting
shell restricting cells to bind while at the same time allowing proteins to dif-
fuse into the actual adsorbent. The disadvantage of fluid distribution, e.g., by
means of horizontal flow restrictors – acting as filter and thus leading to an
accumulation of cell debris and flow heterogeneities – has severely hampered
industrial applications. It was further found that biomass-adsorbent interac-
tions strongly influenced the hydrodynamic situation in the fluidized bed. Fi-
nally, high salt loads as normally observed after cultivation, asked for either
uneconomical dilution of the feed stream or a new type of ligand being able
to operate under such conditions (the terminology of these ligands is still
diverse, e.g., mixed mode, hydrophobic charge induction, multimodal). In
the meantime, the above problems have been solved by a novel design of a
scaleable fluid distribution system [36], the establishment of guidelines for
rational process design and scale up [37, 38], as well as the introduction of
salt-tolerant ligand structures [39, 40], making EBA an industrially used inte-
grated recovery step. Table 2 summarizes principle characteristics of chro-
matographic adsorbents applicable for EBA processes.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 113

5.3.2. High-gradient magnetic separation In HGMS micron size, nonporous


magnetic particles are employed as adsorbents, which by design are devoid of
pore di¤usion and can be separated in magnetic gradients, allowing the col-
lection of the small particles by nonmechanical techniques [41–43]. The tech-
nique combines the ease of a rapid and scaleable batch adsorption step, fol-
lowed by a high-speed recovery enabled by magnetic fields. The separation of
product capture and solid–liquid separation makes this technique more ro-
bust than EBA; however, it does not reach the resolution generally seen with
a chromatographic bed. A first comparison of both techniques can be found
in Hubbuch et al. [44]. Prior to a successful application in industry, only a
few hurdles seem to be left, namely the development of inexpensive and well-
characterized magnetic adsorbent particles of various functionalities as well
as equipment allowing the quantitative recovery of magnetic adsorbents after
the filter capture and the operation under containment or sterile conditions.

5.3.3. Membrane adsorber Membrane adsorbers rely on convective trans-


port in the pores of a microfiltration membrane and can be operated with pre-
clarified feedstreams [45, 46]. When considering membrane chromatography,
the surface of the membrane is modified with a suitable ligand for product cap-
ture. Taken this features together, membrane adsorbers meet the main criteria
for a successful initial step, where a rapid processing of high volumetric and di-
lute feedstreams is required [47]. In addition, it could be shown that membrane
adsorbers exhibit a much higher DNA-binding capacity than found for chro-
matographic beads and thus are used increasingly for DNA removal steps dur-
ing the purification process [48, 49]. Membrane applications thus cover the
whole range, from an initial step to final purification (see below) [50–52].

6. Inclusion body processing and protein folding


Inclusion bodies form during recombinant protein synthesis in the host cell if
the biosynthesis rate exceeds the protein folding rate. Under these conditions,
hydrophobic folding intermediates accumulate and form aggregates, which
may be visible in the cells as refractive bodies using a light microscope. Inclu-
sion bodies form especially in E. coli expressing proteins that contain disul-
fide linkages in the active conformation, because of the reductive environ-
ment in the bacterial cytoplasm. The inclusion bodies contain more or less
of host cell proteins but represent nevertheless an enriched form of product
albeit inactive. As pointed out above, inclusion bodies can be purified by
repeated centrifugation exploiting their higher density [16, 53, 54]. In a now
classic experiment, Anfinsen demonstrated in 1973 using ribonuclease as an
example that the information for a correct folding of a random polypeptide
chain into the active conformation is somehow embedded in the primary

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


114 J. Hubbuch and M.-R. Kula

sequence of the protein [55]. This so-called ‘‘folding code’’ has been exten-
sively studied for many years, but is still not completely understood although
much progress has been made. It should be noted that the active conforma-
tion of a protein represents a small di¤erence between entropic and enthalpic
contributions. A folding pathway exists from the random coil to the final
three-dimensional structures. The first task to transform an inclusion body
into active product is a dissolution of the protein aggregates using chaotropic
agents such as urea or guanidinium salts under reducing and if necessary
slightly alkaline conditions. During renaturation of the protein, the chaot-
ropic agent has to be removed and at the same time re-aggregation from fold-
ing intermediates avoided. Aggregation is concentration dependent; there-
fore, to reach high activity yields a very high dilution is necessary, which is
an expensive operation. Better performance can be obtained by analyzing the
kinetics of the single stages along the pathway. Refolding, e.g., of human tis-
sue plasminogen activator from inclusion bodies has been achieved in large
scale matching the flow rate into a stirred tank to the kinetics of the reaction
and accumulating a much higher protein concentration in solution than by
simple dilution [56]. Other approaches to minimize aggregation are the bind-
ing of the target molecule to solid support during removal of the chaotrop
or the addition of hydrophilic polymers such as polyethylene glycol during
refolding [57, 58]. Correct disulfide linkages are formed mainly by disulfide
exchange reactions, selecting a proper redox environment and including oxi-
dized and reduced forms of glutathione in the refolding bu¤er [59].
Protein folding under native and technical conditions still requires better
understanding for improved control. Perturbation of the active structure upon
binding of proteins to surfaces is another important topic in this context, and
only recently analytical methods have been developed to approach such ques-
tions. Proteins are polyelectrolytes and it is recognized from recent studies in
ultrafiltration that the counter ions may influence conformation as well; de-
tailed understanding is however lacking. These gaps in basic understanding
have to be closed in the future in order to improve process economics.
Proteins successfully renatured from inclusion bodies can be processed further
as discussed below. Aggregates obtained during the refolding step may be re-
cycled to increase the overall yield [60].

7. Product concentration and initial purification

7.1. General aspects


The clarified feed solution obtained by various steps described above contains
the product of interest usually in low or moderate concentration. Removal

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 115

of water is an important part of further processing and expensive. Again,


di¤erent options exist to achieve a concentration and initial purification of
the product. Depending on the nature of the product, the following tech-
niques are commonly employed: extraction, crystallization, precipitation, or
ultrafiltration.

Extraction is well established in the chemical industry and has been briefly
introduced above. For bulk products of low molecular weight, extraction
into organic solvents is increasingly important for environmental as well as
economical considerations.

7.2. Crystallization
Crystallization is established in the biotech industry dealing with steroids,
antibiotics, or amino acids. The technology exploits the equilibrium in the
chemical potential between a crystalline product and its saturated solution,
which can be manipulated by temperature, pH, as well as solvent composi-
tion. The latter may be a¤ected by adding salts or water miscible organic
solvents. Crystals exhibit a highly ordered structure and are formed from
oversaturated solutions if nuclei grow in an ordered fashion to large size. Nu-
cleation and crystal growth can be influenced and steered by seeding tech-
niques, which are widely applied.

In case amorphous aggregates are formed during similar manipulations, the


process is called precipitation; the boundary is not always clear when dealing
with proteins. For decades, the crystallization of proteins was considered an
art and extremely di‰cult. This notion resulted from work aiming at x-ray
di¤raction of proteins to high resolution, which necessitates flawless single
crystals. But if crystallization is employed as a concentration and purification
step, demands on the quality of the single crystal are not so stringent. Glu-
cose isomerase was the first bulk protein crystallized in large scale in industry.
Another prominent example where protein crystallization is used for protein
recovery and purification is the proteo-hormon insulin [61]. Because of re-
combinant production technologies, concentration and purity of extracellular
bulk enzymes in the culture broth are now much higher than previously, and
today crystallization at an early stage is increasingly used in industrial pro-
cesses with or without preceding ultrafiltration for such products [62, 63].
The technology has considerable potential also for the processing of pharma
proteins and may result in the replacement of one or more expensive chroma-
tography steps [64]. To reach this goal, the knowledge base concerning
protein–protein interactions and crystallization from more contaminated
feeds needs much improvement. In principle, the theoretical framework for

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


116 J. Hubbuch and M.-R. Kula

crystallization of small or large molecules is the same, but di¤usion rates and
therefore kinetics are quite di¤erent. Besides, interaction parameters become
much more important for macromolecules such as proteins. In addition, the
stability of individual proteins limits reaction conditions, as proteins can be
deactivated by solvents, heat, or extreme pH values. Nevertheless, large-scale
protein crystallization is gaining momentum as process development ap-
proaches issues ranging from high-throughput screening, the influence of con-
taminants, scale-up and industrial equipment design, as well as robust and
simple models [63, 65, 66].

7.3. Precipitation
Fifty years ago, precipitation was one of a few available methods to purify
proteins and was extensively studied in the development of the plasma frac-
tionation process by Cohn and coworkers (1946) to produce human albumin
[67]. Until today, human albumin is purified in large scale from donated
blood after the removal of blood cells by a series of precipitation steps alter-
ing pH and ethanol concentration during the process and controlling the tem-
perature within narrow limits. Precipitation of bulk enzymes using salts has
been largely replaced by ultrafiltration for environmental reasons. Never-
theless, a better understanding and deeper insight into protein solubility
behavior [68] is necessary, both to reduce product losses during production
as well as for the development of new and industrially feasible procedures.
Good reviews on current approaches on protein precipitation can be found
in [69, 70].

7.4. Ultrafiltration
Modern filter media cover a wide range of pore sizes from nanometers to tens
of micrometers. Proper selection allows separation by the hydrodynamic ra-
dius of species, which correlates with molecular weight for globular proteins.
Ultrafiltration is extensively used for concentration of proteins as well as
for bu¤er exchange; in both cases, the size di¤erence between retained and
passed species is very large. Recent developments show the importance of dif-
fusive transport for protein separation by molecular filtration and the impor-
tance of charges on the membrane as well as the protein for transport. Cer-
tainly an increased understanding of interaction coe‰cients between like and
unlike proteins, as well as with the membrane surface and the bu¤er ions is
needed for further advancement of ultrafiltration as a separation tool. For
the concentration of products with lower molecular weight than proteins,
nanofiltration can be employed utilizing membranes with even smaller pore
sizes [71–73].

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 117

8. Purification

8.1. General aspects


The demand for purity of biotechnological products di¤ers widely and de-
pends on the use of such products. Often repeated crystallization may be suf-
ficient to meet demands. Intravenously administered pharma proteins require
highest purity which can be achieved only by multistage operations such as
chromatography or electrophoresis. Electrophoresis in various forms is indis-
pensable in analytics and small-scale isolation of, e.g., proteins or plasmids,
but has not found large-scale application.

8.2. Chromatography
In industrial-scale production, chromatography is the method of choice. The
largest chromatographic process operated today is the separation of glucose
and fructose by ligand exchange chromatography using ion exchangers
charged with calcium ions and operated continuously in simulating moving
bed mode in the production of HFCS (high fructose corn syrup), annual
production > 106 tons. This example demonstrates that in principle chroma-
tography can be operated in very large scale and even for cheap products!
But this process has several specific features aiding the development: the mol-
ecules are small and have su‰ciently high di¤usivities, water is the mobile
phase and unspecific adsorption is negligible, allowing the utilization of me-
chanically robust, cheap resins. Antibiotics, steroids, and other low molecular
weight products are also purified by chromatography on ion exchange or
reverse phase resins but in a much smaller scale compared to the HFCS
process.

In contrast, chromatographic separation of proteins and even larger particu-


lates is much more di‰cult on a large scale due to technical issues and in
many cases contributes over-proportionally to the overall purification costs.

Chromatography is a complex and dynamic process, where the properties of


the proteins as well as the properties of the resin are important and mass
transport often is rate limiting. Figure 5 shows a detailed picture of the pro-
cesses influencing the overall performance at di¤erent scales, while typical
resin characteristics are given in Table 2. The governing processes on a single
particle (mm-scale) in respect to fluid dynamics are film and pore di¤usion
and the partitioning behavior of the soluble compounds. In addition to these
transport parameters, processes occurring on a nm-scale basis – protein–
protein and protein–adsorbent interactions – complete the picture of a single
adsorbent particle system. The overall operation of a packed chromatography

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


118 J. Hubbuch and M.-R. Kula

Figure 5 Processes governing chromatographic performance at di¤erent scales.

column (cm to m-scale) superimposes processes such as convective flow, dis-


persion, and di¤usion acting in macroscale.
As pointed out above, the chromatographic resin is a porous particle com-
posed mainly of synthetic or natural polymers and produced in bead form
[34]. An overview on the variety of adsorbent architecture including di¤erent
backbone materials, pore structures, and ligands is given in Table 2. In order
to avoid nonspecific binding of proteins and to ensure biocompatibility, hy-
drophilic materials are normally the materials of choice for resin production.
Hydrophilicity and thus biocompatibility can be correlated to the Gibbs free
energy of the solubility for the respective materials. As an example, Dextran
150 000 with a DGm of P41.2 is a clearly hydrophilic material, while Poly-
styrol with a DGm of P23 expresses hydrophobic characteristics. Key pa-
rameters for packed bed applications are a low nonspecific binding, control-
lable pore structure during manufacturing, and a certain rigidity to withstand
pressure. Pore structure and the introduction of ligands into the adsorbent
have proven a very important parameter for resin optimization. Generally
one distinguishes between the pore size, enabling either di¤usive or convec-
tive flow and product-related pore sizes, which provide the optimal adsorbent
load for a product of given size. In recent years, special features such as the
introduction of polymer chains or gels into the pore volume have led to yet
unexplained features such as improved uptake kinetics, higher capacity and
selectivity of the respective resins.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 119

The intraparticle processes are governed by di¤usion and adsorption. This


simple picture, however, is becoming more complex when considering the dy-
namic change of pore accessibility by adsorbed protein and thus a continuous
change of transport controlling boundary conditions as well as competitive
adsorption processes occurring in a multicomponent feed stream. A recently
developed, novel technique that allows the in-situ analysis of intraparticle ad-
sorption patterns and transport processes on the basis of confocal laser scan-
ning microscopy provides a fundamental insight into chromatography and
is expected to act as a basis for the formulation of sophisticated modeling
approaches [74–79]. The ligands are covalently attached to the surface by dif-
ferent spacers; this linkage should be hydrolytically stable to withstand many
cleaning and washing cycles for repetitive use. The spacer employed has
an influence on performance even when the same ligand and the same base
material are used, making resin changes more di‰cult [34]. The contribution
of the resin to separation, however, is little understood today. The highest se-
lectivity can be expected using biospecific ligands in a‰nity chromatography.
Even though they also contribute highest to the cost of goods, they might act
as a platform for processes aiming at the purification of the respective protein
class and thus o¤er enormous savings in process development. The most
prominent example in today’s industry is the use of protein A chromatogra-
phy for the purification of monoclonal antibodies. Table 2 summarizes
the principle characteristics of adsorbents used for protein purification; each
ligand class exploits a more or less specific property of the protein for
interaction.
In general, chromatographic theory is well developed and treated extensively
in specialized monographs. In the absence of adsorptive interactions – e.g., in
gel permeation chromatography – performance usually can be well predicted
taking film and pore di¤usion into account and the apparent di¤usivity as
experimentally determined. For adsorption chromatography, an additional
isotherm model that describes the adsorption behavior in equilibrium state is
required and has to be determined experimentally. For this purpose, either
empirical (e.g., Langmuir) or physically based (e.g., steric-mass-action [80])
isotherm models can be applied. Especially for protein separation by means
of ion-exchange chromatography, the steric mass action model (SMA) has
been proven to be a robust and convenient model [81–83]. Separation can be
accomplished by isocratic or gradient elution or by displacement [84]. Com-
mercial columns are preferentially operated under high load conditions, when
displacement may occur and protein–protein interaction complicate perfor-
mance and predictions. The basic understanding and modeling of processes is
best advanced for ion exchange chromatography and less well for other types
of adsorption chromatography. Modeling assumes dilute solutions and a
single species, both assumptions clearly violated in industrial conditions. But

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


120 J. Hubbuch and M.-R. Kula

validated models containing lumped parameters are nevertheless important


to reduce the number of experiments in scale up, and presently much e¤ort
is invested in this direction. An overview about various model approaches
in chromatography, their complexity, as well as field of application, can be
found in [85].

Traditionally, the most common processing scheme in chromatography is


based on batch processing. In the last decade, however, a novel method based
on continuous counter-current processing is gaining more and more accep-
tance. Compared to the common batch chromatography, chromatographic
separation applying a continuous, counter-current operation mode (e.g., si-
mulated moving bed, SMB) generally yields higher productivity and lower
eluent consumption due to the better utilization of adsorbent in counter-
current mode. In SMB processes, a serial arrangement of 6–24 columns is
used and the counter-current flow between mobile and stationary phase is ‘‘si-
mulated’’ by switching inlet and outlet ports simultaneously in the direction
of mobile phase flow. In doing so, two product fractions can be obtained in
which either the stronger or weaker adsorbed component is enriched. Coming
from the small molecule side, SMB processes have recently been successfully
realized for the separation of biomolecules [82, 86–88].

9. Polishing

9.1. Impurity removal


During the complete purification process but mainly in the last stage in a pu-
rification train trace impurities have to be removed, this may be host cell pro-
teins [89, 90], endotoxins [91, 92], DNA [93–96], viruses [97, 98] or even trun-
cated, clipped or chemically modified forms of the target molecule [71]. Often
membrane adsorbers are employed in flow through mode, but removal of
modified versions of the target molecule usually is very di‰cult and requires
very sophisticated chromatography.

9.2. Final concentration and formulation


Purified products are stored either in the crystalline state, as dried powders,
or in concentrated solutions. Drying has to be gentle enough to maintain the
activity and integrity of the product. Well-established drying technologies
such as spray driers are employed. For proteins, freeze drying is used if the
product is stored and delivered in single vials. Often in the final step additives
are added, which are needed or desired for the final application or to improve
stability of the product and shelf life.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 121

10. Conclusion
Today, pharma proteins represent the fastest growing sector in biotechnol-
ogy. The downstream processing of biotechnological products is a complex,
multiparameter task involving specialized knowledge, highly sophisticated
machinery, as well as high-tech materials. In view of the high risk connected
to a failure in purity and identity of the respective drug and thus stringent
regulations, however, novel purification approaches or newly developed unit
operations take a long time before they are implemented. The greatest di‰-
culties presently encountered are in the ever increasing scale necessary
to meet the demands and the task to produce highly complex and pure mole-
cules under economic limitations. In order to develop new approaches to
meet these goals, a deeper insight into the basic foundations of separation
processes is definitely needed. On the other hand, industrial timelines often
lead to a situation where process understanding and optimization is sacrificed
for the need to be first on the market. Therefore, a strong focus in the biotech
industry is currently placed on the combination of high-throughput experi-
mentation and the application of a sophisticated mathematical optimization
algorithm in order to rapidly evaluate optimal conditions for a given purifica-
tion task [99].

References
[1] Sanger, F., Thompson, E.O.P., The amino-acid sequence in the glycyl chain of
insulin. 1. The investigation of lower peptides from partial hydrolysates, Bio-
chem. J., 53 (1953), 353–366.
[2] Wheelwright, S.M., Protein Purification: Design and Scale up of Downstream
Processing, Hanser Publishers, Munich 1991.
[3] Hejnas, K., Matthiesen, F., Skriver, L., Protein stability in downstream pro-
cessing, in: Bioseparation and Bioprocessing Vol. II, Ed. G. Subramanian,
Wiley-VCH, Weinheim, 1998, pp. 31–61.
[4] Middelberg, A.P.J., The release of intra-cellular bioproducts, in: Bioseparation
and Bioprocessing Vol. II, Ed. G. Subramanian, Wiley VCH, Weinheim, 1998,
pp. 131–162.
[5] Chisti, Y., Moo-Young, M., Disruption of microbial cells for intracellular
products, Enzyme Microb. Technol., 8 (1986), 194–204.
[6] Birnboim, H.C., Doly, J., Rapid alkaline extraction procedure for screening in
recombinant plasmid DNA, Nucleic Acids Res., 7 (1979), 1513–1523.
[7] Prazeres, D.M.F., Ferreira, G.N.M., Monteiro, G.A., Cooney, C.L., Cabral,
J.M.S., Large-scale production of pharmaceutical-grade plasmid DNA for
gene therapy: problems and bottlenecks, Trends Biotechnol., 17 (1999),
169–174.
[8] Urthaler, J., Buchinger, W., Necina, R., Industrial scale cGMP purification of
pharmaceutical grade plasmid-DNA, Chem. Eng. Technol., 28 (2005), 1408–
1420.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


122 J. Hubbuch and M.-R. Kula

[9] Frerix, A., Müller, M., Kula, M.R., Hubbuch, J., Scalable recovery of plasmid
DNA based on aqueous two-phase separations, Biotechnol. Appl. Biochem.,
42 (2005), 57–66.
[10] Levy, M.S., Collins, I.J., Yim, S., Ward, J.M., Titchener-Hooker, N., Sham-
lou, P.A., E¤ect of shear on plasmid DNA in solution, Bioprocess Eng., 29
(1999), 7–13.
[11] Chamsart, S.H., Patel, H., Hanak, J.A.J., Hitchcock, A.G., Nienow, A.W.,
The impact of fluid-dynamic-generated stresses on cDNA and pDNA stability
during alkaline cell lysis for gene therapy products, Biotechnol. Bioeng., 75
(2001), 387–392.
[12] Lin, D.Q., Brixius, P.J., Hubbuch, J.J., Thömmes, J., Kula, M.-R., Biomass/
adsorbent electrostatic interactions in expanded bed adsorption: A zeta poten-
tial study, Biotechnol. Bioeng., 83 (2003), 149–157.
[13] van Hee, P., Middelberg, A.P.J., van der Lans, R.G.J.M., van der Wielen,
L.A.M., Relation between cell disruption conditions, cell debris particle size,
and inclusion body release, Biotechnol. Bioeng., 88 (2004), 100–110.
[14] Hubbuch, J.J., Brixius, P., Lin, D.-Q., Möllerup, I., Kula, M.-K., The influence
of homogenisation conditions on biomass-adsorbent interactions during ion-
exchange expanded bed adsorption, Biotechnol. Bioeng., 94 (2006), 543–553.
[15] Karau, A., Boldt, K., Buchholz, S., Aspekte der Interaktion von modernen
Fermentationstechnologien und Aufarbeitungstechniken bei der Produktion
von Biokatalysatoren, GVC/Dechema Symposium, Downstream Processing/
Separation of Bioproducts, Bad Honef, Germany, May 7, 2002.
[16] Middelberg, A.P.J., O’Neill, B.K., Harvesting recombinant protein inclusion
bodies, in: Bioseparation and Bioprocessing Vol. II, Ed. G. Subramanian, Wi-
ley VCH, Weinheim, 1998, pp. 81–105.
[17] Voisard, D., Meuwly, F., Ru‰eux, P.A., Baer, G., Kadouri, A., Potential of
cell retention techniques for large-scale high-density perfusion culture of sus-
pended mammalian cells, Biotechnol. Bioeng., 82 (2003), 751–765.
[18] Darcy, H., Les Fontaines Publiques de la Ville de Dijon, p. 647, Dalmont,
Paris, 1856.
[19] van Reis, R., Brake, J.M., Charkoudian, J., Burns, D.B., Zydney, A.L., High-
performance tangential flow filtration using charged membranes, J. Membr.
Sci., 159 (1999), 133–142.
[20] Zydney, A.L., van Reis, R., High-performance tangential-flow filtration, Bio-
technol. Bioproc., 26 (2001), 277–298.
[21] Schügerl, K., Hubbuch, J., Integrated bioprocesses. Curr. Opin. Microbiol., 8
(2005), 294–300.
[22] Rü¤er, N., Heidersdorf, U., Kretzers, I., Sprenger, G.A., Raeven, L., Takors,
R., Fully integrated L-phenylalanine separation and concentration using
reactive-extraction with liquid–liquid centrifuges in a fed-batch process with
E. coli, Bioproc. Biosys. Eng., 26 (2004), 239–248.
[23] Albertsson, P.A., Partition of Cell Particles and Macromolecules, Wiley, New
York, 1985.
[24] Hustedt, H., Kroner, K.H., Kula, M.R., Applications of phase partitioning in
biotechnology, in: Partitioning in Aqueous Two Phase Systems, Theory, Meth-
ods, Uses and Applications to Biotechnology, Eds. H. Walter, D.E. Brooks, D.
Fisher, pp. 529–587, Academic Press, New York, 1985.
[25] Rämsch, C., Kleinelanghorst, L.B., Knieps, E.A., Thommes, J., Kula, M.-R.,
Aqueous two-phase systems containing urea: Influence on phase separation

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 123

and stabilization of protein conformation by phase components, Biotechnol.


Progr., 15 (1999), 493–499.
[26] Rämsch, C., Kleinelanghorst, L.B., Knieps, E.A., Thommes, J., Kula, M.-R.,
Aqueous two-phase systems containing urea: Influence of protein structure on
protein partitioning, Biotechnol. Bioeng., 69 (2000), 83–90.
[27] Frerix, A., Schönewald, M., Geilenkirchen, P., Müller, M., Kula, M.-R.,
Hubbuch, J., Exploitation of coil-globule pDNA transition induced by small
changes in temperature, pH salt and PEG compositions for directed partition-
ing in aqueous two phase systems, Langmuir, 22 (2006), 4282–4290.
[28] Kula, M.-R., Selber, K., Protein purification by aqueous liquid–liquid extrac-
tion, in: Encyclopedia of Bioprocess Technology, Eds. M.C. Flickinger, S.W.
Drew, pp. 2179–2191, John Wiley & Sons, New York, 1999.
[29] Selber, K., Tjerneld, F., Collen, A., Hyytia, T., Nakari-Setala, T., Bailey, M.,
Fagerstrom, R., Kan, J., van der Laan, J., Penttila, M., Kula, M.-R., Large-
scale separation and production of engineered proteins, designed for facilitated
recovery in detergent-based aqueous two-phase extraction systems, Proc. Bio-
chem., 39 (2004), 889–896.
[30] Papamichael, N., Hustedt, H., Enzyme recovery by continuous crosscurrent
extraction, Methods Enzymol., 228 (1994), 573–584.
[31] Thömmes, J., Halfar, M., Gieren, H., Curvers, S., Takors, R., Brunschier, R.,
Kula, M.-R., Human chymotrypsinogen B production from Pichia pastoris by
integrated development of fermentation and downstream processing, Part II:
Protein recovery, Biotechnol. Progr., 15 (2001), 503–512.
[32] Selber, K., Nellen, F., Ste¤en, B., Thommes, J., Kula, M.-R., Investigation of
mathematical methods for e‰cient optimisation of aqueous two-phase extrac-
tion, J. Chromatogr. B, 743 (2000), 21–30.
[33] Bensch, M., Selbach, B., Hubbuch, J., High throughput screening techniques
in downstream processing: Preparation, characterization and optimisation of
aqueous two phase systems, Chem. Eng. Sci. (2007), in press.
[34] Müller, E., Properties and characterization of high capacity resins for biochro-
matography, Chem. Engin. Technol., 28 (2005), 1295–1305.
[35] Thömmes, J., Fluidized bed adsorption as a primary recovery step in protein
purification, Adv. Biochem. Eng. Biotechnol., 58 (1997), 185–230.
[36] Hubbuch, J.J., Heeboll-Nielsen, A., Hobley, T.J., Thomas, O.R.T., A new
fluid distribution system for scale-flexible expanded bed adsorption, Biotech-
nol. Bioeng., 78 (2002), 35–43.
[37] Lin, D.Q., Fernandez-Lahore, H.M., Kula, M.R., Thömmes, J., Minimizing
biomass/adsorbent interactions in expanded bed adsorption process: A meth-
odological approach, Bioseparation, 10 (2001), 7–19.
[38] Hubbuch, J.J., Thömmes, J., Kula, M.-R., Expanded bed adsorption –
biochemical engineering aspects. Adv. Biochem. Eng. Biotechnol., 92 (2005),
101–123.
[39] Hamilton, G.E., Luechau, F., Burton, S.C., Lyddiatt, A., Development of a
mixed mode adsorption process for the direct product sequestration of an extrac-
ellular protease from microbial batch cultures, J. Biotechnol., 79 (2000), 103–115.
[40] Burton, S.C., Harding, D.R.K., Salt-independent adsorption chromatography:
New broad-spectrum a‰nity methods for protein capture, J. Biochem. Bio-
phys. Methods, 49 (2001), 275–287.
[41] Hubbuch, J.J., Thomas, O.R.T., High gradient magnetic a‰nity separation of
trypsin from porcine pancreatin, Biotechnol. Bioeng., 79 (2002), 302–312.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


124 J. Hubbuch and M.-R. Kula

[42] Heebøll-Nielsen, A., Dalkiær, M., Hubbuch, J.J., Thomas, O.R.T., Super-
paramagnetic adsorbents for high-gradient magnetic fishing of lectins out of
legume extracts, Biotechnol. Bioeng., 87 (2004), 311–323.
[43] Franzreb, M., Siemann-Herzberg, M., Hobley, T.J., Thomas, O.R.T., Protein
purification using magnetic adsorbent particles, Appl. Microbiol. Biotechnol.,
70 (2006), 505–516.
[44] Hubbuch, J.J., Matthiesen, D.B., Hobley, T.J., Thomas, O.R.T., High gradi-
ent magnetic separation versus expanded bed adsorption: A first principle com-
parison, Bioseparation, 10 (2001), 99–102.
[45] Gebauer, K., Thömmes, J., Kula, M.-R., Plasma protein fractionation with
advanced membrane adsorbents, Biotechnol. Bioeng., 54 (1997), 181–189.
[46] Gebauer, K., Thömmes, J., Kula, M.-R., Breakthrough-performance of high-
capacity membrane adsorbers in protein chromatography, Chem. Eng. Sci., 52
(1997), 405–419.
[47] Gottschalk, U., Downstream processing of monoclonal antibodies: From high
dilution to high purity, BioPharm International, June (2005), 42–58.
[48] Gottschalk, U., Fischer-Fruehholz, S., Reif, O., Membrane adsorbers. A cut-
ting edge process technology at the threshold, BioProcess International, May
(2004), 56–65.
[49] Han, B., Specht, R., Wickramasinghe, S.R., Carlson, J.O., Binding Aedes ae-
gypti densonucleosis virus to ion exchange membranes, J. Chromatogr. A, 1092
(2005), 114–124.
[50] Thömmes, J., Kula, M.-R., Membrane chromatography – an integrative con-
cept in the downstream processing of proteins, Biotechnol. Progress, 11 (1995),
357–367.
[51] van Reis, R., Zydney, A., Membrane separations in biotechnology, Curr.
Opin. Biotechnol., 12 (2001), 208–211.
[52] Gosh, R., Review: Protein separation using membrane chromatography, op-
portunities and challenges, J. Chromatogr. A, 952 (2002), 13–27.
[53] Middelberg, A.R., Preparative protein refolding, Trends Biotechnol., 20 (2002),
437–443.
[54] Tsumoto, K., Ejima, D., Kumagai, I., Practical considerations in refolding
proteins from inclusion bodies, Protein Expr. Purif., 28 (2003), 1–8.
[55] Anfinsen, C.B., Principles that govern the folding of protein chains, Science,
181 (1973), 223–230.
[56] Misawa, S., Kumagai, I., Refolding of therapeutic proteins produced in
Escherichia coli as inclusion bodies, Biopolymers, 51 (1999), 297–307.
[57] Jungbauer, A., Kaar, W., Schlegl, R., Folding and refolding of proteins in
chromatographic beds, Curr. Opin. Biotechnol., 15 (2004), 487–494.
[58] Cleland, J.L., Builder, S.E., Swartz, J.R., et al. Polyethylene-glycol enhanced
protein refolding, Biotechnology, 10 (1992), 1013–1019.
[59] Bulaj, G., Formation of disulfide bonds in proteins and peptides, Biotechnol.
Adv., 23 (2005), 87–92.
[60] Machold, C., Schlegl, R., Buchinger, W., et al., Continuous matrix assisted refold-
ing of alpha-lactalbumin by ion exchange chromatography with recycling of ag-
gregates combined with ultradiafiltration, J. Chromatogr. A, 1080 (2005), 29–42.
[61] Brage, J., Galenics of Insulin, Springer, Berlin, 1987.
[62] Jacobsen, C., Garside, J., Hoare, M., Nucleation and growth of microbial li-
pase crystals from clarified concentrated fermentation broths, Biotechnol. Bio-
eng., 57 (1998), 666–674.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 125

[63] Klyushnichenko, V., Protein crystallization: From HTS to kilogram-scale,


Curr. Opin. Drug. Discov. Devel., 6 (2003), 848–854.
[64] Peters, J., Minuth, T., Schröder, W., Implementation of a crystallization step
into the purification process of a recombinant protein, Protein Expr. Purif., 39
(2005), 43–53.
[65] Lee, E.K., Kim, W.S., Protein crystallization for large-scale bioseparation,
Biotechnol. Bioproc., 27 (2003), 277–320.
[66] Carbone, M.N., Judge, R.A., Etzel, M.R., Evaluation of a model for seeded
isothermal batch protein crystallization, Biotechnol Bioeng., 91 (2005), 84–90.
[67] Cohn, E.J., Strong, L.E., Hughes Jr, W.L., Mulford, D.J., Ashworth, J.N.,
Melin, M., Taylor, H.L., Preparation and properties of serum and plasma pro-
teins. IV. A system for the separation into fractions of the protein and lipopro-
tein components of biological tissue and fluids, J. Am. Chem. Soc., 68 (1946),
459–475.
[68] Prausnitz, J.M., Molecular thermodynamics for some applications in biotech-
nology, Pure Appl. Chem., 75 (2003), 859–873.
[69] Kumar, A., Galaev, I.Y., Mattiasson, B., Precipitation of proteins: Nonspecific
and specific, Biotechnol. Bioproc., 27 (2003), 225–275.
[70] Hilbrig, F., Freitag, R., Protein purification by a‰nity precipitation, J. Chro-
matogr. B, 790 (2003), 79–90.
[71] Ebersold, M.F., Zydney, A.L., The e¤ect of membrane properties on the sepa-
ration of protein charge variants using ultrafiltration, J. Membr. Sci., 243
(2004), 379–388.
[72] Shao, J.H., Zydney, A.L., Retention of small charged impurities during ultra-
filtration. Biotechnol. Bioeng., 87 (2004), 7–13.
[73] Baruah, G.L., Venkiteshwaran, A., Belfort, G., Global model for optimizing
crossflow microfiltration and ultrafiltration processes: A new predictive and de-
sign tool. Biotechnol. Progr., 21 (2005), 1013–1025.
[74] Hubbuch, J.J., Linden, T., Knieps, E., Thömmes, J., Kula, M.-R., Dynamics
of protein adsorption within the adsorbent particle during packed bed chroma-
tography, Biotechnol. Bioeng., 80 (2002), 359–368.
[75] Hubbuch, J.J., Linden, T., Knieps, E., Ljunglöf, A., Thömmes, J., Kula, M.-
R., Mechanism and dynamics of protein transport in chromatographic media
studied by confocal laser scanning microscopy. Part 1: The interplay of sorbent
structure and fluid phase conditions, J. Chromatogr. A, 1021 (2003), 93–104.
[76] Hubbuch, J.J., Linden, T., Knieps, E., Thömmes, J., Kula, M.-R., Mechanism
and dynamics of protein transport in chromatographic media studied by con-
focal laser scanning microscopy. Part 2: Impact on chromatographic separa-
tions, J. Chromatogr. A, 1021 (2003), 105–115.
[77] Schröder, M., von Lieres, E., Hubbuch, J., Direct quantification of intra-
particle protein di¤usion in chromatographic media, J. Phys. Chem. B, 110
(2006), 1429–1436.
[78] Teske, C.A., Schroeder, M., Simon, R., Hubbuch, J.J., Protein labelling e¤ects
in confocal laser scanning microscopy, J. Phys. Chem. B, 109 (2005), 13811–
13817.
[79] Teske, C.A., von Lieres, E., Schröder, M., Ladiwala, A., Cramer, S., Hub-
buch, J., Competitive adsorption of labeled and native protein in confocal laser
scanning microscopy, Biotechnol. Bioeng., 95 (2007), 58–66.
[80] Brooks, C.A., Cramer, S.M., Steric-mass-action ion exchange: Displacement
profiles and induced salt gradients, AIChE J., 38 (1992), 1969–1978.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


126 J. Hubbuch and M.-R. Kula

[81] Iyer, H., Tapper, S., Lester, P., Wolk, B., van Reis, R., Use of the steric mass
action model in ion-exchange chromatographic process development, J. Chro-
matogr. A, 832 (1999), 1–9.
[82] Wekenborg, K., Susanto, A., Frederiksen, S., Schmidt-Traub, H., Nicht-
isokratische SMB-Trennung von Proteinen, Chem. Ing. Tech., 75 (2003),
1081.
[83] Pedersen, L., Mollerup, J., Hansen, E., Jungbauer, A., Whey proteins as a
model system for chromatographic separation of proteins, J. Chrom. B, 790
(2003), 161.
[84] Seidel-Morgenstern, A., Preparative gradient chromatography, Chem. Eng.
Technol., 28 (2005), 1265–1273.
[85] Schmidt-Traub, H. (Ed.), Preparative Chromatography of Fine Chemicals and
Pharmaceutical Agents, Wiley-VCH, Weinheim, 2005.
[86] Beltscheva, D., Hugo, P., Seidel-Morgenstern, A., Linear two-step gradient
counter-current chromatography analysis based on a recursive solution of an
equilibrium stage model, J. Chrom. A, 989 (2003), 31–45.
[87] Houwing, J., van Hateren, S.H., Billiet, H.A., van der Wielen, L.A.M., E¤ect
of salt gradients on the separation of dilute mixtures of proteins by ion-
exchange in simulated moving beds, J. Chrom. A, 952 (2002), 85–98.
[88] Paredes, G., Makart, S., Stadler, J., Mazzotti, M., Simulated moving bed
operation for size exclusion plasmid purification, Chem. Engin. Technol., 28
(2005), 1335–1345.
[89] Graumann, K., Ebenbichler, A.A., Development and scale up of preparative
HIC for the purification of a recombinant therapeutic protein, Chem. Eng.
Technol., 28 (2005), 1398–1407.
[90] Yigzaw, Y., Piper, R., Tran, M., Shukla, A.A., Exploitation of the adsorptive
properties of depth filters for host cell protein removal during monoclonal an-
tibody purification, Biotechnol. Prog., 22 (2006), 288–296.
[91] Montbriand, P.M., Malone, R.W., Improved method for the removal of endo-
toxin from DNA, J. Biotechnol., 44 (1996), 43–46.
[92] Karplus, T.E., Ulevitch, R.J., Wilson, C.B., A new method for reduction of
endotoxin contamination from protein solutions, J. Immunol. Methods, 105
(1987), 211–220.
[93] Christensen, P.A., Danielczyk, A., Stahn, R., Goletz, S., Simple separation of
DNA in antibody, Protein Expr. Purif., 37 (2004), 468–471.
[94] Giovannini, R., Freitag, R., Isolation of a recombinant antibody from cell cul-
ture supernatant: Continuous annular versus batch and expanded-bed chroma-
tography. Biotechnol. Bioeng., 73 (2001), 522–529.
[95] Levy, M.S., Collins, I.J., Tsai, J.T., Shamlou, P.A., Ward, J.M., Dunnill, P.,
Removal of contaminant nucleic acids by nitrocellulose filtration during
pharmaceutical-grade plasmid DNA processing, J. Biotechnol., 76 (2000),
197–205.
[96] Persson, J., Nystrom, L., Ageland, H., Tjerneld, F., Purification of recombi-
nant proteins using thermoseparating aqueous two-phase system and polymer
recycling. J. Chem. Technol. Biotechnol., 74 (1999), 238–243.
[97] Kuriyel, R., Zydney, A.L., Sterile filtration and virus filtration, Methods Bio-
technol., 9 (2000), 185–194.
[98] Bohonak, D.M., Zydney, A.L., Compaction and permeability e¤ects with vi-
rus filtration membranes, J. Membr. Sci., 254 (2005), 71–79.

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2


Isolation and Purification of Biotechnological Products 127

[99] Bensch, M., Schulze Wierling, P., von Lieres, E., Hubbuch, J., High through-
put screening of chromatographic phases for rapid process development, Chem.
Eng. Technol., 28 (2005), 1274–1284.
[100] Rushton, A., Ward, A.S., Holdich, R.G., Solid–liquid filtration and separation
technology, Wiley VCH, Weinheim, 1996.

Paper received: 2006-04-19


Paper accepted: 2006-07-12

J. Non-Equilib. Thermodyn.  2007  Vol. 32  No. 2

You might also like