488 Biotechnol. Prog.

2008, 24, 488−495

REVIEWS
Advances in Primary Recovery: Centrifugation and Membrane Technology
David J. Roush*,† and Yuefeng Lu‡
BioPurification Development, Merck Research Laboratories, P.O. Box 2000, Mailstop RY805S-100, Rahway, New Jersey 07065,
and Process Development, Amgen, One Amgen Center Drive, Mailstop 30W-2-A, Newbury Park, California 91320

Significant and continual improvements in upstream processing for biologics have resulted in
challenges for downstream processing, both primary recovery and purification (1). Given the
high cell densities achievable in both microbial and mammalian cell culture processes, primary
recovery can be a significant bottleneck in both clinical and commercial manufacturing. The
combination of increased product titer and low viability leads to significant relative increases in
the levels of process impurities such as lipids, intracellular proteins and nucleic acid versus the
product. In addition, cell culture media components such as soy and yeast hydrolysates have
been widely applied to achieve the cell culture densities needed for higher titers (2, 3). Many of
the process impurities can be negatively charged at harvest pH and can form colloids during the
cell culture and harvest processes. The wide size distribution of these particles and the potential
for additional particles to be generated by shear forces within a centrifuge may result in insufficient
clarification to prevent fouling of subsequent filters. The other residual process impurities can
lead to precipitation and increased turbidity during processing and even interference with the
performance of the capturing chromatographic step. Primary recovery also poses significant
challenges owing to the necessity to execute in an expedient manner to minimize both product
degradation and bioburden concerns. Both microfiltration and centrifugation coupled with depth
filtration have been employed successfully as primary recovery processing steps. Advances in
the design and application of membrane technology for microfiltration and dead-end filtration
have contributed to significant improvements in process performance and integration, in some
cases allowing for a combination of multiple unit operations in a given step. Although these
advances have increased productivity and reliability, the net result is that optimization of primary
recovery processes has become substantially more complicated. Ironically, the application of
classical chemical engineering approaches to overcome issues in primary recovery and purification
(e.g., turbidity and trace impurity removal) are just recently gaining attention (4). Some of these
techniques (e.g., membrane cascades, pretreatment, precipitation, and the use of affinity tags)
are now seen almost as disruptive technologies (5). This paper will review the current and potential
future state of research on primary recovery, including relevant papers presented at the 234th
American Chemical Society (ACS) National Meeting in Boston.

Introduction Few examples (8) exist for the simultaneous optimization of

Advancements in primary recovery have been driven by both
experimental and modeling approaches (6). Ongoing improve-
ments in primary recovery steps are required to match the
increases in productivity and cell densities achieved via
continued cell culture development (7). Significant research has
been devoted to the development of scale-down models and
scale-up strategies for microfiltration. However, a representative
scale-down model for disk-stack centrifugation that incorporates
all of the functionality of the pilot and production scale (e.g.,
partial discharge), mimics the potential shear damage that can
be incurred in the flow path and can be executed with limited
amounts of developmental material (<30 L) remains a challenge.
* To whom correspondence should be addressed. Ph: (732) 594-3204.
Email: david_roush@merck.com.
† Merck Research Laboratories.
‡ Amgen.
10.1021/bp070414x CCC: $40.75 © 2008 American Chemical Society and American Institute of Chemical Engineers
Published on Web 04/15/2008
cell culture production (including time of harvest) and primary
recovery (either centrifugation or MF coupled with DF). Hence,
a significant amount of primary recovery process development
and optimization remains empirically driven.
Microfiltration
Microfiltration has historically been utilized as the main
primary recovery step for biologics (9) since it affords a
separation of biomass from the product of interest and can also
be used to achieve significant purification via selective sieving
of product or impurities. Van Reis and Zydney (10) have
recently prepared an excellent review article summarizing the
key developments and utilization of this technology to date,
including the key engineering parameters evaluated during
development. Microfiltration has been employed as the primary
recovery part of a platform process for purification of mono-
Biotechnol. Prog., 2008, Vol. 24, No. 3
clonal antibodies (11), and recent advances in modeling of shear
damage to cells (12) have led to a better understanding of this
unit operation. Development of a microfiltration-based primary
recovery step can be accomplished with minimal feedstock
owing to the availability of small cassettes (0.1 m2 at comparable
path length to full scale devices). However, this technology can
be extremely sensitive to changes in feedstock quality (e.g., cell
culture viability, cell density, medium components). Research
by Huang et al. (13) and Romero et al. (14) on MF suggests
that some of these limitations, including the high transmembrane
pressure (TMP) associated with a combination of high cell
density and low viability, can be overcome via the use of pH
or ionic strength control. Use of cationic polymers as a
flocculating agent to enhance the performance of MF for the
clarification of bacterial cell culture broth has also been proposed
(15). With the advent of low shear centrifuges, the popularity
of microfiltration for the harvest of mammalian cell culture
broths has been significantly reduced.
As mentioned previously, the use of membranes early in the
purification process can afford significant purification. One
logical extension is the use of a series of membranes or a
membrane cascade (16) to afford purification of a solubly
expressed biologic. This approach is essentially a continuum
of membranes with modified feed and flow properties which
exploit differences in sieving coefficients which result in
improved global separation factors. The idea builds on the
concepts recently developed to improve selectivity of ultrafil-
tration (UF) membranes (17-19) including HPTFF (20) (high
performance tangential flow filtration).
Centrifugation
Centrifugation coupled with depth filtration has become the
main set of unit operations employed for the primary recovery
of therapeutic proteins in recent years (21-23). Scale-up of
centrifugation has historically been performed using the Sigma
concept (24) where the goal is to maintain the ratio of the flow
rate (Q) to the equivalent clarification or settling area (Σ), Q/Σ,
as the process scale is increased. However, this methodology
has some significant limitations in that different centrifuge
designs are often employed during scale-up from lab to
commercial production. The differences in design can lead to
dramatic differences in performance as a function of scale (25)
requiring a reduction (up to 70%) in Q/Σ at production scale in
order to achieve the desired clarification efficiency. Another
challenge in the development of a centrifugation process is the
lack of existence of a small scale centrifuge that incorporates
all of the functionality of a large scale unit (e.g., partial discharge
functionality). The smallest currently available unit, the West-
falia CSA-1, has a total bowl volume of 600 mL and a sediment
holding space of 250 mL. A significant amount of material
(typically >10 L) is required to perform a single experiment in
this system.
Many parameters associated with the operation of the
centrifuge (g-force, residence time, discharge frequency) and
the associated cell culture feedstock (e.g., cell density, viability
at harvest) can impact the effective clarification efficiency of
the centrifuge. Although some computational fluid dynamics
(CFD) modeling has been performed to determine the relative
impact of shear on the performance of the centrifugation step
(26), only a limited amount of published data exist on the impact
of these parameters on the performance of the centrifugation
step (27). The result is a significant number of empirical studies
are performed to determine the optimal operating conditions
for a given system. Recently, a more holistic approach to
489
optimization of the primary recovery step has been employed
to account for the impact of cell culture density and conditions
(28) on the performance of the entire primary recovery step
(centrifugation and depth filtration) (29-31).
An ultra-scale-down model has been developed by Boychyn
et al. (32) based on a combination of CFD and experimental
studies. This model incorporates the impact of shear damage
on cells via the use of a small scale shear cell and a lab
centrifuge to approximate performance of a disk stack centrifuge.
Menon et al. (33) employed aspects of this ultra-scale-down
model to evaluate applicability to a mammalian cell culture
process stream. Menon et al. compared the process performance
of the ultra-scale-down model to the performance of a pilot scale
centrifuge (equivalent to a CSA-1) that was equipped with a
low shear inlet zone. Menon et al. first mapped the performance
(turbidity of the centrate) of the pilot scale centrifuge to changes
in Q/Σ over a range from <5.0 × 10-9 to ∼2.0 × 10-8. An
assessment was then performed with the ultra-scale-down model
under conditions where the feed (cell culture broth) did not
experience any pre-shearing in the shear cell. The results
indicated that the pilot scale centrifuge performed equivalently
to the no shear state suggesting minimal shear damage is
incurred during processing through the pilot scale centrifuge.
Performance of the subsequent clarifying filtration (0.2 µm),
assessed via filter loading capacity was also comparable in the
two systems. Additional experiments indicated that increasing
shear rates in the scale-down model by employing the shear
cell produced a significant increase in turbidity and a reduction
in the mean particle size of the resulting centrate.
This study clearly confirmed that if there is a difference in
the shear imparted on the cell culture broth in the centrifuge,
either due to differences in equipment design or operational
parameters of the primary recovery step, a significant change
in separation performance can result. This is consistent with
the previous observations of Hutchison et al. It is important to
note that earlier research by Tebbe et al. (34) also demonstrated
the important impact of equipment design on the separation
performance of a disk stack centrifuge. Specifically, Tebbe et
al. demonstrated that the use of a CSA-1 centrifuge equipped
with a hydrohermetic entry zone, a feature which significantly
reduces the shear forces in the inlet to the centrifuge by
removing the air/liquid interface, results in minimal cell lysis
and associated submicron particle formation versus a standard
(non-hermetic) design. Although it was encouraging to observe
that under some conditions the pilot scale unit performed
comparably to the ultra-scale-down model, it is also important
to note that these were the observations for a given cell culture
system (cell density, viability, etc.) and that the results may
not be transferable to other systems. Hence, additional research
on the performance of a disk-stack centrifuge with other
experimental systems is warranted.
In addition to optimizing the clarification efficiency of the
centrifugation step, another practical aspect is to maximize yield.
Every time a disk-stack centrifuge ejects the solids space there
is a potential to lose some of the associated product bearing
stream (cell culture supernatant or potential centrate). Typical
feed to a disk stack centrifuge contains ∼1 × 107 cells/mL and
hence frequent discharges may be required to minimize the
potential for cell solids breakthrough into the product stream,
an event that would significantly reduce the loading capacity
for the subsequent DF step. Several mechanisms to overcome
this loss were reviewed during the recent ACS session, including
regulation of timing of initiation and termination of flow to the
centrifuge and the use of partial discharges of the bowl. Under
490
the later condition, only a fraction of the bowl is emptied each
time the bowl is opened. This method is often favored since
the disk-stack centrifuge can be rapidly reset and feed reinitiated.
Unfortunately both of the operating modes allow for the
introduction of air into the bowl space resulting in an air-liquid
interface that has the potential to increase the shear on the cells
and degrade the performance of the centrifuge-consistent with
the observations of Menon et al. This degradation in perfor-
mance was also observed by Zhao et al. (35) as higher turbidity
immediately after a partial discharge was executed. This increase
in turbidity had a significantly deleterious effect on the
subsequent depth filtration performance. Better clarification
performance was achieved using full shots coupled with a buffer
flush to minimize the air-liquid interface. In the future, the
issue of partial versus full shots may be overcome via the use
of a new equipment design that allows for continuous discharge
of the solids stream from the centrifuge. The rate of continuous
discharge of solids from the bowl is a strong function of the
outlet nozzle design. The nozzle design would need to be
optimized depending on the phase ratio of the feed stream, with
more restrictive nozzles required for lower solid/liquid ratios
(e.g., mammalian cell culture processes) to avoid excessive loss
of supernatant to the waste stream. Current design constraints
do not allow for this flexibility, thereby limiting this approach
to systems with high cell density. Other researchers (36) have
explored an alternative centrifuge design, a continuous discharge
tubular bowl centrifuge. Excellent clarification efficiency and
yield were obtained with this design for a range of feeds.
Unfortunately, only a small manufacturing base exists for this
technology making scale-up and implementation challenging.
Depth Filtration
Depth filtration (DF) is widely used in the clarification of
cell culture broth prior to capture chromatography, serving as
the secondary clarification step following centrifugation or
tangential flow microfiltration. In certain infrequent cases, depth
filters are used directly to clarify cell culture broth. The most
frequently used depth filters for bioprocessing consist of
diatomaceous earth (DE), perlite, cellulose fibers, and a
positively charged resin binder. Some depth filters also have
hydrophobic interaction characteristics (10, 31, 37, 38).
Depth filters, unlike absolute filters, retain particles throughout
the porous media allowing for retention of particles both larger
and smaller than the pore size. Particle retention is believed to
involve both size exclusion and adsorption through hydrophobic,
ionic and other interactions (Figure 1).
The fouling mechanism may include pore blockage, cake
formation and/or pore constriction (39, 40). The high particulate
load and high turbidity present in clarified cell culture super-
natant adds challenges to the secondary clarification by depth
filtration. If not properly developed or optimized, over-sized
depth filters may be required at production scale resulting in
yield losses and scale-up issues.
In order to improve depth filtration capacity, depth filter
screening at lab scale is generally performed to identify filters
optimal for the intended feed stream (41). Proper pore size grade
selection allows for more effective utilization of the entire depth
of the porous filter media for particle retention instead of just
surface retention. Depth filters composed of multilayer structures
with graded pore size allows removal of particles of different
size at different layers of the filter leading to increased filtration
capacity or improved filtration efficiency (at a fixed filter
loading). A two-stage depth filtration scheme composed of
multilayer depth filters over a range of porosities provides
Biotechnol. Prog., 2008, Vol. 24, No. 3
Figure 1. Mechanisms of particle retention by depth filters. Depth
filters retain particles through both retention and adsorption. Proper
grade selection results in effective use of the depth of the filter media
and not just the surface. Improper selection results in only surface
filtration, which leads to short filter life. (Figure adapted from CUNO,
a 3M Company, with permission)
broader coverage for a range of particle size distributions. This
approach was successfully applied to clarify centrifugation pools
of very high turbidities (42). A significant advance in the
technology would be a depth filter with deep gradient-density
to provide a wide span of pore size grades.
Owing to their charge and hydrophobic characteristics, depth
filters have been applied to remove endotoxin, DNA, host cell
proteins and potential adventitious agents such as viral particles
and prions. Depth filters have been used for the removal of
endotoxin from water (43) and DNA from cell culture super-
natant (44). Positively charged depth filters have also been used
to reduce host cell protein through ionic and hydrophobic
interactions and prevent the fouling of the subsequent capture
column (31).
The charge capacity of depth filters may be assessed by
measuring the DNA content in the filtrate. DNA is very
negatively charged and is able to interact with depth filter media
under physiological pH and ionic strength. Thus it can serve as
an intrinsic marker for measuring charge capacity (45). The
capacity for DNA removal in this case was exhausted after
loading about 100 L/m2 of centrifuged cell culture broth (Figure
2).
The binding capacities of depth filters for DNA and CHO
proteins (CHOP) have also been measured directly using cell
culture fluid post microfiltration clarification (41). In this case
study, the depth filters were able to adsorb 3.9-7.6 g/m2 of
DNA and 0.4-5.4 g/m2 of CHOP. The adsorption capacity of
the filters may have been saturated at the test conditions since
the reduction of DNA and CHOP were 25-44% and 9-17%,
respectively, relative to the level in the load. While removal of
DNA from cell culture fluid appeared to be significant although
incomplete, the removal of CHOP was found to be limited at
physiological pH and conductivity, similar to that of membrane
chromatography (10).
Another application of depth filtration is clarification of the
product pool post capture chromatography. When depth filter
is applied in this application it may provide more meaningful
reduction for the trace or very low levels of host cell proteins
(HCP) and DNA in the pools (46). Solamo et al. demonstrated
that the extent of reduction was affected by the pH, conductivity
and/or protein concentration of the load (Figure 3A and B).
Through nonspecific interaction with the capturing Protein
A affinity resin or through ionic and/or hydrophobic interactions
Biotechnol. Prog., 2008, Vol. 24, No. 3
Figure 2. Comparison of DNA clearance for two commercially
available depth filters as a function of loading. Depth filtration with
Millipore A1HC, 5.4 m2 and CUNO 90ZA, 0.3 m2 filters was carried
out with a fed batch mammalian cell culture broth post centrifugation.
Flow rate was set at 200 L/m2/h. The load had a DNA level of
approximately 2.85E+08 pg/mL. Results obtained by Roman et al. (45).
Figure 3. Reduction of HCP in post low pH viral inactivation pool
by depth filtration as a function of flux. The degree of reduction was
affected by pH of the load (A). In addition, diluting the load with water
to lower the conductivity and product concentration also increased HCP
reduction (B). Similar observations were made with DNA reduction
by Solamo et al. (46).
with the antibody molecules, the increased levels of negatively
charged host cell proteins, nucleic acids, lipids, and media
components in modern cell culture fluids may survive both
clarification and affinity purification and be carried over into
the capturing column pool. At the relatively low pH of 4 to 6
in the Protein A pool and/or subsequent post low-pH viral
inactivation pool, host cell impurities and/or media components
may be acid precipitated out of the solution, resulting in turbid
pools. Highly turbid Protein A column pools and/or post low-
pH viral inactivation pools have becoming increasingly prob-
lematic. An extremely large size normal flow (0.2 µm) filter
491
would typically be required to clarify the pool prior to
subsequent chromatographic purification. Using depth filter in
conjunction with normal flow 0.2 µm filters led to significantly
reduced membrane filter requirements (47). Similar to the
performance of DF in primary recovery, removal of host cell
impurities (DNA and HCP) was observed. Since the product
concentrations are much higher in the post viral-inactivation
pool, depth filtration devices with lower hold-up volumes, such
as the POD format from Millipore (48), may provide higher
step yield without significantly diluting the product.
In addition to clearance of host cell impurities and particu-
lates, removal of both retrovirus and parvo virus by depth filters
has also been evaluated (49, 50). In general, the depth filters
were able to achieve one to greater than five logs of viral
reduction under varied testing conditions. Tipton et al. has
suggested that viral clearance, similar to the purification afforded
by DF for process impurities, involves both ionic and hydro-
phobic interactions. Owing to the inherent filter product
variability and multiple particulate removal mechanisms in-
volved, the development and demonstration of a DF step for
viral clearance is extremely challenging.
Scale-Down Models and Scale-Up Considerations
Small size depth filter disks are often used for the purpose
of initial screening to compare different brands of depth filters
to fit particular feed streams (51). Variations of up to 30% in
filter performance of small filters are expected because of the
wet-laid filter manufacturing process. Hence, it is recommended
that multiple lots of the filters need to be tested in order to
achieve an accurate assessment of filtration performance. Small
size disk depth filter performance evaluations cannot provide
reliable filter scaling-up information in terms of filter uniformity,
geometry structure, flow distribution, and hold-up volume.
Observations from small scale experiments will need to be
confirmed at pilot scale before further scaling up to production.
If the supply of feed material is not limited, representative small
scale depth filters from 0.1 to 1.0 m2 of size may be more
reliable for filter selection and to be used for centrifugation unit
op development. However, representative feed streams from the
first stage of primary recovery (either centrifugation or micro-
filtration) must be employed for these experiments. A qualified
scale-down model is needed during process characterization
(52). A scale-down model that can be validated is typically not
needed for depth filtration. Integrity testing of the depth filter
is also typically not required during the production process.
Overall Optimization of Primary Recovery Processes
Including Disruptive Technologies
Both Menon and Zhao reviewed in detail the importance of
optimizing the entire primary recovery process, including the
use of precipitation based pretreatment. This research builds
on previous examples (53, 54) of the use of precipitating agents,
sometimes charged directly to the bioreactor (2, 55) to improve
the performance of the subsequent primary recovery steps (both
CE and DF).
Flocculation and acid precipitation have been explored to
improve clarification efficiency, yield and/or robustness during
the primary recovery from mammalian cell culture broth (41,
42, 56, 57). Both technologies have been used extensively in
wastewater treatment and food and chemical industries (58).
The majority of flocculants used are positively charged. Some
examples are polyamines, cationic polysaccharides, chitosan,
and diallyldimethyl ammonium chloride. The positive charged
polymers may act by bridging the negatively charged particles
492
Figure 4. Mechanisms of flocculation between a charged polyelec-
trolyte and oppositely charged particle. Note: shown here is an anionic
(negatively charged) solid interacting with a polycationic (positively
charged) polymer. Adapted from Prof. Todd Menkhaus’ research
webpage (http://webpages.sdsmt.edu/∼tmenkhau/research.htm) with
permission.
together to form large particles (Figure 4). Alternatively, they
may neutralize the charges of the small particles to allow them
to agglomerate to form large particles (isoelectric precipitation)
(59).
One recent example on the use of flocculation was a
precipitation step using calcium phosphate formed in the cell
culture broth via addition of calcium chloride and potassium
phosphate (2). The underlying mechanism was thought to be
related to coprecipitation of calcium phosphate with cell culture
debris and impurities, with the positively charged calcium
interacting with the negatively charged cell debris and process
impurities such as DNA, lipids, and host cell proteins. Acid
precipitation may be carried out using simple inorganic and
organic acids such as phosphoric acid, sulfuric acid, acetic acid,
and citric acid (57, 42, 60). The mechanism for the acid
precipitation may be related to lower solubility of intracellular
proteins at acidic pH, and cellular DNA may be precipitated
because of complexes formed with chromatin (57). The solubil-
ity of certain culture medium components may also be pH-
dependent.
Flocculation was shown to dramatically improve clarification
performance. For example (56), chitosan used as a flocculant
in mammalian cell culture dramatically improved centrifugation
as indicated by reduced centrate turbidity and increased
clarification filter capacity. The flocculation process was robust
as it performed well in a relatively wide range of pH (5-7.5),
conductivity (e340 mM NaCl) and chitosan concentration
(0.002-0.005%). Product quality and clarification yield were
not affected. Chitosan is a polymer of deacetylated chitin, which
is polycationic at acidic and neutral pH, is not animal sourced,
is generally considered safe, and has been widely used in devices
and in pharmaceutical preparation (61). One challenge that
remains is demonstrating clearance by subsequent process steps.
In another case study, flocculation was carried out with calcium
phosphate (2). Calcium phosphate flocculation also provided
significantly improved centrifugation and subsequent filtration
performance. Product yield from the calcium phosphate floc-
culation was typically acceptable but variability still existed.
Acid precipitation was carried out on centrate derived from
a hybridoma cell culture broth (57). The yield of the subsequent
UF/DF step was improved, though any impact on the throughput
was not discussed. In another case study (42), acid precipitation
was performed directly in the bioreactor allowing for removal
Biotechnol. Prog., 2008, Vol. 24, No. 3
Figure 5. Impact of acid precipitation (AP) on the turbidity of
downstream post low pH viral-inactivation pools. Approximately a
3-fold drop in turbidity was observed post AP. Cell culture broth was
titrated to pH of approximately 5.0 with acetic acid, clarified by
centrifugation and depth filtration and loaded on to the Protein A
column. The Protein A column pools went through low pH viral
inactivation and titrated to pH of about 5.0. The turbidity of the post
viral inactivation pools was measured by OD600nm (42).
of the precipitate with the cells in the subsequent centrifugation
and depth filtration step. The ultimate impact was an increase
in capacity of the depth filtration at the same flux by more than
3 fold as shown in a lab scale depth filtration testing. In both
case studies it is important to note that product quality was not
impacted based on biochemical and/or activity analysis. In a
third case study (60), acid precipitation was carried out prior to
microfiltration or centrifugation (62). The flux of the microfil-
tration operation was dramatically improved. The yield of the
microfiltration was also improved from <83% to >90%.
Incorporating acid precipitation prior to centrifugation also
improved depth filtration capacity by ∼10-fold. One advantage
of acid precipitation over the flocculation is its relative simple
process. In the case of the cell culture clarification and
monoclonal antibody purification, there is no need to show
clearance of simple acids such as acetic acid or sulfuric acid.
Mixing conditions need to be carefully evaluated if acid
precipitation is employed to minimize the potential for locally
low pH gradients resulting in product degradation, especially
if strong acids are employed.
One added benefit of the flocculation and acid precipitation
is the potential to remove process impurities (2, 57, 60) such
as host cell impurities and DNA resulting in improved perfor-
mance of the downstream purification steps. Flocculation with
calcium phosphate significantly reduced the turbidity of the
Protein A peak. Acid precipitation of the centrate from the
hybridoma cell culture broth eliminated the precipitation
otherwise observed in the UF/DF pool during its storage. Acid
precipitation also significantly reduced the turbidity of the
Protein A pool or post low-pH viral inactivation pool (Figure
5) (42).
Chitosan flocculation did not have significant effects on the
performance of the capturing Protein A affinity chromatography
step in terms of step yield and HCP removal (33). The impact
on Protein A pool or post low pH viral-inactivation pool
turbidity was not discussed but could conceivably be similar to
that of acid precipitation and calcium flocculation as the chitosan
flocculation process likely also reduced the levels of negatively
charged process impurities.
Flocculation and acid precipitation have also been applied
in E. coli processes. Karim et al. (54) evaluated using floccula-
tion to improve centrifugation and microfiltration of E. coli cell
lysate. Acid precipitation, on the other hand, has been frequently
Biotechnol. Prog., 2008, Vol. 24, No. 3
used in the clarification during the post-refolding harvest of E.
coli-expressed proteins (62). In addition, polyelectrolyte floc-
culation and subsequent flocculant removal to enhance solids
clarification has also been evaluated for their potential applica-
tion in biofuel process streams or recovery of recombinant
proteins from transgenic agriculture. In summary, flocculation
and acid precipitation have been demonstrated to be effective
approaches in enhancing the clarification processes for the
primary recovery of biotech products.
The ultimate extension of this technique would be the
development of a completely non-chromatographic separation
process composed of primary recovery coupled with precipita-
tion (either product specific, impurity specific or a combination
of both). An entire session at the last two ACS National
Meetings has been devoted explicitly to the topic of non-
chromatography separations. One excellent example presented
at the Boston ACS meeting by McDonald et al. (63) was the
use of polyelectrolytes (polyvinyl sulfonic acid polymer) to
replace either protein A or cation-exchange (CEX) chromatog-
raphy. This future state will require an even better fundamental
theoretical understanding and optimization of the performance
of the primary recovery unit operations in order to develop a
robust downstream process. Without this additional research,
the performance of the subsequent purification steps will
continue to be subject to significant variability owing to changes
in performance of the bioreactor and primary recovery steps
(64).
Conclusions
Despite the excellent progress to date on development of
scale-down models and methodologies for primary recovery unit
operations, ongoing fundamental research (including CFD
modeling) to optimize the combination of CE + DF is
warranted. The utility of an ultra-scale-down model to evaluate
the impact of shear sensitivity to primary recovery performance
has been proven. However, with the advent of low shear
centrifuges, shear damage (as measured via increases in turbidity
or host cell components) is not a significant issue under most
operating conditions but may become an issue upon scale-up.
A significant amount of empirical development is still required
owing to the large number of parameters which can impact the
performance of the primary recovery step and subsequent unit
operations. Significant potential exists for the application of
precipitation-based methodologies to both enhance the perfor-
mance of existing primary recovery processes and potentially
replace downstream chromatography steps.
Acknowledgment
The authors wish to thank Kent Göklen (Merck), Anurag
Rathore (Amgen), Xiaoyang Zhao (Amgen), and Nihal Tugcu
(Merck) for insightful suggestions for the manuscript.
References and Notes
(1) Low, D.; O’Leary, R.; Pujar, N. Future of antibody purification, J.
Chromatogr. B 2007, 848, 48-63.
(2) Shpritzer, R.; Vivik, S.; Orlando, S.; Acharya, H.; Coffman, J.
Calcium phosphate flocculation of antibody-producing mammalian
cells at pilot scale. 232nd American Chemical Society National
Meeting, September 10-14, 2006, San Francisco, CA; BIOT
division, Abstract 80.
(3) Chun, B.-H.; Kim, J.-H.; Lee, H.-J.; Chung. N. Usability of size-
excluded fractions of soy protein hydrolysates for growth and
viability of Chinese hamster ovary cells in protein-free suspension
culture. Bioresour. Technol. 2007, 98 (5), 1000-1005.
(4) Thommes, J.; Etzel, M. Alternatives for chromatographic separa-
tions. Biotechnol. Prog. 2007, 23, 42-45.
493
(5) Przybycien, T.; Pujar, N.; Steele, L. Alternative bioseparation
operations: life beyond packed-bed chromatography. Curr. Opin.
Biotechnol. 2004, 15, 469-478.
(6) Velayudhan, A.; Menon, M. Modeling of purification operations
in biotechnology: Enabling process development, optimization and
scale-up. Biotechnol. Prog. 2007, 23, 68-73.
(7) DePalma, A. Designing better purification schemes. Genet. Eng.
News 2007, 27, 44-47.
(8) Alexander, C.; Shamlou, P.; Breen, L.; Mostafa, S. Development
of a robust process for harvesting bioreactors. 2006 American
Institute of Chemical Engineers Meeting, San Francisco, CA,
November 2006; Abstract 457d.
(9) Russotti, G.; Göklen, K. Crossflow Membrane Filtration of
Fermentation Broth, Membrane Separations in Biotechnology, 2nd
ed.; Marcel Dekker, Inc.: New York, 2000; Chapter 3, pp 85-159.
(10) Van Reis, R.; Zydney, A. Bioprocess membrane technology. J.
Membr. Sci. 2007, 297, 16-50.
(11) Parola, S.; Tugcu, N.; Roush, D. J. Evolution and scale-up of
microfiltration in antibody production. 229th American Chemical
Society Meeting, San Diego, March 2005; BIOT Division, Abstract
4.
(12) Vickroy, B.; Lorenz, K.; Kelly, W. Modeling shear damage to
suspended CHO cells during cross-flow filtration, Biotechnol. Prog.
2007, 23, 194-199.
(13) Huang, P.; Peterson, J.; Jorgensen, J.; Petersen, B.; Duffy, B.;
Wesson, M.; Zhou, W. Membrane applications in development of
humanized antibody manufacturing processes. Advanced Membrane
Technology II Conference, May 23-28, 2004, Irsee, Germany.
(14) Romero, J.; Chrostowski, J.; de Vilmorin, P.; Case, J. Effects of
pH and ionic conditions on microfiltration of mammalian cells:
Combined permeate flux enhancement and mAb purification capa-
bilities. 2006 American Institute of Chemical Engineers Annual
Meeting, San Francisco, CA, November 2006; Extended Abstract
195A.
(15) Aspelund, M.; Glatz, C.; Graves, K.; Rozeboom, G.; Heng, M.
Improving permeate flux in the microfiltration of a bacterial cell
suspension by flocculation with cationic polyelectrolytes. 2006
American Institute of Chemical Engineers Annual Meeting, San
Francisco, CA, November 2006; Abstract 195b.
(16) Lightfoot, E. N.; O’Dell, J. L. Ideal membrane cascades for
downstream processing. 234th American Chemical Society Meeting,
Boston, MA, August 2007; BIOT Division, Abstract 40.
(17) Ebersold, M.; Zydney, A. The effect of membrane properties on
the separation of protein charge variants using ultrafiltration, J.
Membr. Sci. 2004, 379-388.
(18) Zydney, A. Application of charged ultrafiltration membranes for
bioprocessing, 229th American Chemical Society Meeting, San
Diego, CA, March 2005; BIOT Division, Abstract 1.
(19) Lightfoot, E. Can membrane cascade replace chromatography?
Sep. Sci. Technol. 2005, 40, 740-756.
(20) Van Reis, R.; Zydney, A. Membrane separations in biotechnology,
Curr. Opin. Biotechnol. 2001, 12, 208-211.
(21) Kempken, R.; Preissman, A.; Berthold, W. Assessment of a disc
stack centrifuge for use in mammalian cell separation. Biotechnol.
Bioeng. 1995, 46 (2), 132-138.
(22) Pham, C.; Lee, M.; Westoby, M.; Thömmes, J. Solid-liquid
separation of mammalian cell culture suspensions using a continuous
disk stack centrifuge. 3rd International IBC Recovery and Purification
Conference, November 18, 2002, San Diego, CA, IBC Life Sciences,
Westborough, MA.
(23) Shpritzer, R. Evaluation of a continuous disc stack centrifuge for
the clarification of mammalian cell cultures. 225th American
Chemical Society National Meeting, March 2003, New Orleans, LA.
(24) Maybury, J.; Hoare, M.; Dunnill, P. The use of laboratory
centrifugation studies to predict performance of industrial ma-
chines: Studies of shear-insensitive and shear-sensitive materials,
Biotechnol. Bioeng. 2000, 67 (3), 265-273.
494
(25) Hutchison, N.; Bingham, N.; Murrell, N.; Farid, S.; Hoare, M.
Shear stress analysis of mammalian cell suspensions for prediction
of industrial centrifugation and its verification. Biotechnol. Bioeng.
2006, 95 (3) 483-491.
(26) Boychyn, M.; Yim, S.; Shamlou, P.; Bulmer, M.; More, J.; Hoare,
M. Characterization of flow intensity in continuous centrifuges for
the development of laboratory mimics. Chem. Eng. Science 2001,
56 (16), 4759-4770.
(27) Salte, H.; King, J.; Tichener-Hooker, N. Centrifugation selection
for recovery of high solids density cell broths and visualization of
separation performance via Windows of Operation. 229th American
Chemical Society Meeting, San Diego, CA, March 2005; BIOT
Division, Abstract 8.
(28) Pampel, L.; Hart, R.; Leite, U.; Zhao, X. Innocent until proven
guilty? Scale-dependent impact of cell culture on recovery perfor-
mance. Recovery of Biological Products XII, April 2006, Litchfield,
AZ.
(29) Iammarino, M.; Nti-Gyabaah, J.; Chandler, M.; Roush, D. Impact
of cell density and viability on primary clarification of mammalian
cell broth. 232nd American Chemical Society Meeting, San Fran-
cisco, September 2006; BIOT Division, Abstract 303.
(30) Iammarino, M.; Nti-Gyabaah, J.; Chandler, M.; Roush, D.; Göklen,
K. Impact of cell density and viability on primary clarification of
mammalian cell broth using disc stack centrifuge and charged depth
filtration. BioProcess Int. 2007, Nov, 38-50.
(31) Yigzaw, Y.; Piper, R.; Tran, M.; Shukla, A. Exploitation of the
adsorptive properties of depth filters for host cell protein removal
during monoclonal antibody purification. Biotechnol. Prog. 2006,
22 (1), 288-296.
(32) Bochyn, M.; Yim, S; Bulmer, M.; More, J.; Bracewell, D; Hoare,
M. Performance prediction of industrial centrifuges using scale-down
models, Bioprocess Biosyst. Eng. 2004, 26 (6), 385-391.
(33) Menon, M.; Hamilton, A.; Riske, F. Optimization of cell culture
harvest clarification using an ultrascale-down model for centrifuga-
tion. 234th American Chemical Society Meeting, Boston, MA,
August 2007; BIOT Division, Abstract 41.
(34) Tebbe, H.; Lutkemeyer, D.; Gudermann, F.; Heidemann, R.;
Lehmann, J. Lysis free separation of hybridoma cells by continuous
disk stack centrifugation, Cytotechnology 1996, 22, 119-127.
(35) Zhao, Z.; Zhou, J. Effect of solid ejection size on continuous
centrifugation during mammalian cell culture product recovery. 234th
American Chemical Society Meeting, Boston, MA, August 2007;
BIOT Division, Abstract 42.
(36) Lander, R.; Daniels, C; Meacle, F. Efficient, Scalable Clarification
of Diverse Bioprocess Streams using A Novel Pilot Scale Tubular
Bowl Centrifuge, BioProcess Int. 2005, Nov, 32-40.
(37) Ireland, T.; Bolton, G.; Noguchi, M. Optimizing virus filter
performance with prefiltration, BioProcess Int. 2005, 3 (10)S, 44-
47.
(38) Zhou, J. X.; Solamo, F.; Hong, T.; Shear, M.; Tressel, T. Viral
clearance using disposable systems in mAb commercial downstream
processing. Biotechnol. Bioeng. 2007, submitted for publication.
(39) Laska, M. E.; Brooks, R. P.; Gayton, M.; Pujar, N. S. Robust
scale-up of dead end filtration: impact of filter fouling mechanisms
and flow distribution, Biotechnol. Bioeng. 2005, 92, 308-20.
(40) Wang, A.; Lewus, R.; Rathore, A. S. Comparison of different
options for harvest of a therapeutic protein product from high cell
density yeast fermentation broth, Biotechnol. Bioeng. 2006, 94 (1),
91-104.
(41) Kandula, J. R.; Jain, S.; Ma, J.; Schilling, B. M.; Shukla, A.; Lee,
S. S. Optimization and implementation of harvest depth filtration
for a fusion protein from mammalian cell culture-A case study.
234th American Chemical Society Meeting, August 2007, Boston,
MA; BIOT Division, Abstract 43.
(42) Roman, B.; Zhao, X.; Lu, Y. Development of a robust clarification
process for mAb purification. BioProcess International 2005 Confer-
ence & Exhibition, September 19-22, 2005, Boston, MA.
(43) Gerba, C. P.; Hou, K. Endotoxin removal by charge-modified
filters, Appl. EnViron. Microbiol. 1985, 50 (6), 1375-1377.
(44) Charlton, H. R.; Relton, J. M.; Slater, N. K. Characterization of
a generic monoclonal antibody harvesting system for adsorption of
DNA by depth filters and various membranes, Bioseparation 1999,
8 (6), 281-291.
Biotechnol. Prog., 2008, Vol. 24, No. 3
(45) Roman, B.; Solamo, F.; Tressel, T.; Lu, Y. Development of a
depth filtration step for a recombinant monoclonal antibody. The
Waterside Conference 9th Annual Meeting of Monoclonal &
Recombinant Antibody, May 2-5, 2004; Poster.
(46) Solamo, F.; Zhou, J. X.; Dermawan, S.; Hong, T.; Shear, M.;
Tressel, T. Utilization of depth filters for removal of non-product
related impurities in a large scale monoclonal antibody. Purification
of Biological Products, 1st Annual Meeting, December 5-7, 2005,
Santa Monica, CA.
(47) Lu, Y.; Solamo, F.; Roman, B. Challenge issues in antibody
purification, 232th American Chemical Society National Meeting,
September, 2006, San Francisco, CA; BIOT Division, Abstract
415.
(48) Yavorsky, D. Pod filter design enables integrity testing of large
area depth filters, Next Generation Pharm. (US ed) 2005, December,
Issue 4.
(49) Zhou, J. X.; Solamo, F.; Hong, T.; Shear, M.; Tressel, T. Viral
clearance using disposable systems in mAb commercial downstream
processing. Biotechnol. Bioeng. 2007, submitted for publication.
(50) Tipton, B.; Boose, J. A.; Larsen, W.; Beck, J.; O’Brian, T.
Retrovirus and parvovirus clearance from an affinity column product
using adsorptive depth filtration. BioPharm 2002, September, 43-
50.
(51) Wang, A.; Lewus, R.; Rathore, A. S. Comparison of different
options for harvest of a therapeutic protein product from high cell
density yeast fermentation broth. Biotechnol. Bioeng. 2006, 94 (1),
91-104.
(52) Rathore, A. S.; Wang, A. Optimization, scale up, and validation
issues in filtration of biopharmaceuticals, Part 1. Biopharm Int. 2004,
August.
(53) Duffy, R.; Chan, A.; Balsara, T.; Partridge, B.; Benson, J.; Zhong,
S.; Camera, H.; Mikkelsen, J.; Zhou, W.; Huang, P. Improving
primary cell separation and clarification of humanized monoclonal
antibodies from high yield cell cultures using cationic polymers.
229th American Chemical Society Meeting, San Diego, CA, March
2005; BIOT Division, Abstract 34.
(54) Karim, N.; Han, B.; Graham, H.; Karra, S. Flocculation enhanced
centrifugation and microfiltration of Escherichia coli lysate. 2006
American Institute of Chemical Engineers Annual Meeting, San
Francisco, CA, November 2006; Abstract 438i.
(55) Coffman, J.; Shpritzer, R.; Vicik, S. Flocculation of antibody
producing mammalian cells with precipitating solutions of soluble
cations and anions. Recovery of Biological Products XII, April 2-7,
2006, Litchfield, AZ.
(56) Riske, F.; Schroeder, J.; Belliveau, J.; Kang, X.; Kutzko, J.;
Menon, M. K. The use of chitosan as a flocculant in mammalian
cell culture dramatically improves clarification throughput without
adversely impacting monoclonal antibody recovery, J. Biotechnol.
2007, 128 (4), 813-823.
(57) Lydersen, B. K.; Brehm-Gibson, T.; Murel, A. Acid precipitation
of mammalian cell fermentation broth. Ann. N.Y. Acad. Sci. 1994,
745, 222-231.
(58) Bolto, B.; Gregory, J. Organic polyelectrolytes in water treatment.
Water Res. 2007, 41 (11), 2301-2324.
(59) Hofland, G. W.; de Rijke, A.; Thiering, R.; van der Wielen, L.
A.; Witkamp, G. J. Isoelectric precipitation of soybean protein using
carbon dioxide as a volatile acid. J. Chromatogr. B, Biomed. Sci.
Appl. 2000, 743 (1-2), 357-368.
(60) Huang, P.; Wesson, M.; Zhou, W. Development of a low pH cell
removal and its scale up to 10KL scale for large-scale antibody
manufacturing. BioProcess International Conference and Exhibition,
October 1-4, 2007, Boston, MA.
(61) Özbas-Turan, S.; Akbuga, J.; Aral, C. Controlled release of
interleukin-2 from chitosan microspheres, J. Pharm. Sci. 2002, 91
(5), 1245-1251.
Biotechnol. Prog., 2008, Vol. 24, No. 3
(62) Hart, R. A; Smith, S. G.; Shultz, J. E. Post refolding harvest of
E. coli expressed Fc-fusion proteins. BioProcess International 2005
Conference & Exhibition, September 19-22, 2005, Boston, MA.
(63) McDonald, P.; Carter-Franklin, J.; Fahrner, R. Selective precipita-
tion using polyelectrolytes: A novel approach to the purification of
monoclonal antibodies. 234th American Chemical Society Meeting,
Boston, MA, August 2007; BIOT Division, Abstract 440.
495
(64) Jensen, O.; Staby, A. Change of purification principle in a
recombinant capture process. 229th American Chemical Society
Meeting, San Diego, CA, March 2005; BIOT Division, Abstract 33.
Received October 30, 2007. Accepted March 12, 2008.
BP070414X