Review
Recent advances in large-scale
production of monoclonal antibodies
and related proteins
Abhinav A. Shukla1 and Jörg Thömmes2
1
Manufacturing Sciences & Technology, Bristol-Myers Squibb, Co., 6000 Thompson Road, East Syracuse, NY 13057
2
Technical Development, Biogen Idec, 5200 Research Place, San Diego, CA 92122
The rapid development of high-yielding and robust man-
ufacturing processes for monoclonal antibodies is an
area of significant focus in the biopharmaceutical land-
scape. Advances in mammalian cell culture have taken
titers to beyond the 5 g/l mark. Platform approaches to
downstream process development have become widely
established. Continuous evolution of these platforms is
occurring as experience with a wider range of products is
accrued. The increased cell culture productivity has
shifted the attention of bioprocess development to oper-
ations downstream of the production bioreactor. This
has rejuvenated interest in the use of non-chromato-
graphic separation processes. Here, we review the cur-
rent state-of-the-art industrial production processes,
focusing on downstream technologies, for antibodies
and antibody-related products and discuss future ave-
nues for evolution.
Large-scale production of monoclonal antibodies
Monoclonal antibodies (mAbs) and related proteins have
been shown to successfully target a wide range of extra-
cellular targets with high specificity. This has led to their
introduction to a variety of disease therapies, including
several forms of cancer, multiple sclerosis and immuno-
logical disorders such as rheumatoid arthritis and psor-
iasis (Table 1). Significant advances have been made over
the years in the design of mAbs as therapeutics that have
improved bioavailability, optimized affinity, improved
binding specificity, and human antibody sequences to
reduce any immunogenic side effects [1]. Further advances
in protein engineering are improving the therapeutic pro-
files of mAbs [2–5]. Several cases for which mAb thera-
peutics have been approved serve a large patient
population and involve chronic therapy with high doses.
This has led to the need for production processes that can
be developed rapidly to produce consistently and reprodu-
cibly large quantities of pharmaceutical-grade mAbs at
moderate costs. Process platforms in both upstream cell
culture and downstream purification have become widely
established in industry to meet this need.
Large-scale production of mAbs utilizes mammalian pro-
duction systems followed by cell removal and purification
through sequential chromatographic and membrane
filtration steps, to consistently reduce product- (e.g. protein
Corresponding author: Shukla, A.A. (abhinav.shukla@bms.com).
0167-7799/$ – see front matter ß 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.tibtech.2010.02.001 Available online 19 March 2010
variants) and non-product-related (e.g. host cell proteins)
impurities to acceptable levels. Batch chromatography,
which is traditionally used as the core purification method
in industrial bioprocesses, has long been labeled as a slow
process that requires large volumes of mobile phase and
results in very large pool volumes. Recent advances in the
design of chromatographic stationary phases available for
purification of biomolecules have provided some solutions to
the throughput dilemma, and have facilitated reasonably
productive platforms for industrial protein purification. As
multiple biopharmaceutical companies have become
increasingly involved with the development of mAb pro-
ducts, platform approaches to their production have become
well established in industry. These platforms leverage sim-
ilarities in the biochemical properties and chromatographic
behavior of this product class as well as experience gained
from the first mAbs and Fc fusion proteins that came on the
market. The platforms have enabled significant efficiencies
Glossary
Capture step: the first downstream processing step that captures the product
from the harvested cell broth, concentrates the product, and achieves
separation from impurities that are most unlike the product (e.g. cells, cell
debris, DNA, most proteins).
Cell specific productivity (Qp): productivity of a cell culture expressed in terms
of amount of protein produced per cell per unit time. A conventional unit for
mammalian cell culture is pg/cell/day.
Downstream process: process steps associated with the purification of a
recombinant protein and removal of impurities.
Fed-batch cell culture: a production process based upon feeding a growth
limiting nutrient to the culture. This allows the achievement of a high cell
density in the production bioreactor and facilitates metabolic control of the
cells to avoid generation of side products.
Follow-on biologics (also known as ‘biosimilars’): biopharmaceuticals that are
deemed comparable in quality, safety and efficacy to a reference product from
an innovator company.
NGNA: N-glycolylneuraminic acid that can form the terminal saccharide unit
for an N-linked oligosaccharide structure. Presence of NGNA has been linked to
possible immunogenicity. The usual terminal sugar unit is typically NANA (N-
acetylneuraminic acid).
Polishing steps: steps that occur after the initial capture step that are aimed at
removal of smaller levels of impurities left in the product stream that have
more similarity to the product (e.g. aggregated forms of the product, protein
structural and sequence variants).
Sanitization: procedures that destroy contaminating microorganisms on a
piece of equipment used for bioprocessing. For chromatographic columns,
mild alkaline solutions or high concentrations of chaotropes (e.g. urea) are
used for sanitization to achieve bioburden control while preserving the resin.
Transient transfection: process of introducing DNA into a host cell by non-viral
means such that the genetic material is not inserted into the nuclear genome.
As a result, the genetic material is lost over time through mitosis.
Upstream process: process steps associated with the production of a
recombinant protein by culture and propagation of the host cells.
253
Review Trends in Biotechnology Vol.28 No.5

Table 1. Monoclonal antibodies and related proteins on the market

Name Target Indication Company Year Antibody type
Monoclonal antibodies
Orthoclone OKT3 CD3 Acute kidney transplant rejection Ortho Biotech 1986 Murine
ReoPro Platelet GP IIb/IIIa Prevention of blood clot Centocor 1994 Murine
Rituxan CD20 Non-Hodgkin’s Lymphoma Genentech/ Biogen-Idec 1997 Chimeric
Panorex 17A-1 Colorectal cancer GlaxoSmithKline 1995 Murine
Zenapax IL2Ra (CD25) Acute kidney transplant rejection Hoffman-LaRoche 1997 Humanized
Simulect IL2R Prophylaxis of acute organ rejection Novartis 1998 Chimeric
Synagis RSV Respiratory Synctial Virus Medimmune 1998 Humanized
Remicade TNFa Rheumatoid arthritis Centocor 1998 Chimeric
Herceptin Her2 Metastatic breast cancer Genentech 1998 Humanized
Mylotarg CD33 Acute mylogenous lymphoma Wyeth-Ayerst 2000 Humanixed
Campath CD52 B cell chronic lymphocytic leukemia Takeda 2001 Humanized
Zevalin CD20 Non-Hodgkins Lymphoma Biogen-Idec 2002 Murine
Humira TNFa Rheumatoid arthritis Abbott 2002 Human
Bexxar CD20 Non-Hodgkins Lymphoma Corixa/GSK 2003 Murine
Xolair IgE Allergy Genentech/Novartis 2003 Humanized
Erbitux EGFR/Her1 Colorectal cancer Bristol-Myers 2004 Humanized
Squibb/Imclone (Eli Lilly)
Avastin VEGF Colorectal cancer Genentech 2004 Humanized
Raptiva CD11a Psoriasis Genentech/Xoma 2004 Humanized
Tysabri A4 integrin Multiple sclerosis Biogen-Idec/Elan 2004 Humanized
Vectibix EGFR Colorectal cancer Amgen 2006 Human
Soliris C5 complement PNH – paroxysmal nocturnal Alexion 2007 Humanized
hemoglobinuria
Stelara IL12 and IL23 Psoriasis Centocor 2008 Human
Simponi TNFa Rheumatoid arthritis Centocor 2008 Human
Actemra IL-6 Rheumatoid arthritis Roche 2009 Humanized
Monoclonal antibody fragments
Lucentis VEGF-A Age related macular degeneration Genentech 2006 Fab’
Cimzia TNFa Crohn’s disease UCB 2008 Pegylated Fab fragment
Fc fusion proteins
Enbrel TNFa Rheumatoid arthritis, psoriasis, Amgen 1998 Soluble TNFa receptor
ankylosing spondilitis fused to IgG1 Fc
Amevive CD 2 Psoriasis Biogen-Idec 2003 LFA3 fused to IgG1 Fc
Orencia CD80/86 Rheumatoid arthritis Bristol-Myers Squibb 2005 CTLA4 fused to IgG1 Fc
Arcalyst IL-1 CAPS – Cryopyrin Associated Regeneron 2007 IL-1 receptor fused to Fc
Periodic Syndrome

in process development from time and resource perspect-
ives. Allied areas in biopharmaceutical enterprises, such as
quality and manufacturing, have also benefited through
templating their documents and procedures. A few publi-
cations have provided details on the process platforms in use
at various companies [6–8], and further details have been
shared at conferences. The sections below provide details on
current platforms and review recent developments that are
leading to their evolution. The use of these production plat-
forms has enabled the rapid and efficient introduction of
large numbers of mAbs into clinical trials. The process
platform continues to evolve as further experience is gained
with a wider set of mAbs.
Downstream process platform for mAb production
Upstream process platforms in terms of cell line, clone
selection methodologies, fed-batch cell culture operating
conditions and media have evolved significantly in the last
decade. While these two sections focus on downstream
process platforms, Box 1 provides details about process
operating conditions for large-scale cell culture production
of mAbs.
The downstream purification process has evolved
toward a common framework across industry in recent
years [6–8,28,29]. The platform approach does not imply
a templated process; indeed, this has proven difficult even
within the same company because of differences in proper-
ties and purification behavior between various mAbs [6].
What the platform achieves is the creation of a common
philosophy and alignment over the types of unit operations
to include in the downstream process. Figure 2 shows some
of the downstream process platform schemes adopted at
biopharmaceutical companies involved in mAb process
development. The schematic diagram for a typical down-
stream process, including other filtration and viral-clear-
ance-oriented process steps, is depicted in Figure 3.
All of these schemes rely on the use of Protein A affinity
chromatography as the capture step in the process. This
mode of affinity chromatography is based on the specific
binding affinity between the Fc region of mAbs and the
Protein A ligand. This specificity allows host cell proteins,
DNA and other impurities from the cell culture process to
flow through while the product binds to the stationary
phase, providing >98% purity in a single step. Elution
from Protein A affinity adsorbents is effective under con-
ditions of low pH. In many ways, the high specificity of the
Protein A step has facilitated the adoption of a platform
approach for downstream purification of mAbs. The polish-
ing steps in the downstream process aim to reduce process-
and product-related impurities, particularly host cell

254
Review
Box 1. Upstream process platform
The productivity of fed-batch mammalian cell culture processes for
mAbs has increased significantly in recent years. Two key factors
have underpinned the increase [9]: (i) highly productive cell lines that
have both the right growth characteristics and a high specific
productivity Qp; and (ii) improved understanding of chemically
defined media and feeding strategies to achieve high cell density
and sustained viability over the course of the bioreactor run. Figure 1
is a schematic diagram for an upstream production process that is
often used for the production of mAbs and other glycoproteins.
The creation of a rapid-growth and high-specific-productivity (Qp)
production cell line sets the right foundation for developing the cell
culture production process. Additional requirements include accep-
table cell line stability and the ability to grow in chemically defined
media [10]. The use of effective expression systems [9,10] is a pre-
requisite to creating a high-producing cell line. Commonly utilized
selection markers within the stable cell line development vectors
include those based on genes that encode dihydroxy folate reductase
and glutamine synthetase. These expression vectors also typically
utilize strong gene promoters that drive product mRNA transcription,
such as cytomegalovirus or EF-1a, whereas expression of the gene
that encodes the selectable marker is often driven through a weak
promoter to maximize selection stringency. A high level of product
synthesis in Chinese hamster ovary (CHO) cells has been linked to
increased gene copy number and effective transcription [11]. Loss in
stability of the cell line in the absence of selective pressure has been
linked to decreased transcription rather than loss of copy number,
which is often indicted for instability [10]. Although the complete
molecular mechanisms are incompletely understood, transgene
silencing has been implicated in the decreased productivity. Cell
lines used for antibody expression include the most commonly used
CHO cell lines, murine lymphoid cell lines (e.g. NS0 and SP2/0),
human PER.C6, and often murine hybridoma cell lines [12,13]. All of
these cell lines have been shown to produce product glycosylation
patterns that are compatible with the human immune system,
although there have been reports of high NGNA levels in NS0-
derived expression systems [14]. Cells have also been engineered to
increase product expression through coexpression of important
proteins in the secretory pathway [15]. A number of coexpression
strategies have been researched including coexpression of chaper-
ones, protein degradation inhibitors and apoptosis inhibitors. Obtain-
ing more complete knowledge of the CHO proteome will certainly aid
coexpression efforts to obtain a more productive cell line.
Production cell lines are screened during their development to
select for those with a high specific productivity (Qp). However, not all
cell lines with a high Qp (generally 20–90 pg/cell/day [16]) may
perform well in a production process. As a result of this and other
factors, including stability profiles and desirable product quality
metrics, large numbers of clones need to be generated and screened
to arrive at the best combination of properties that are most suitable
for biopharmaceutical production. Several automated systems for cell
line screening are in use today to facilitate the identification of the
most suitable production line. Fluorescence-activated cell sorting is
one of the techniques used to screen cell lines with high levels of
antibody production from within a heterogeneous population during
selection. A clone screening technique that utilizes relative mRNA
transcript levels to screen predictively for high expressers, using the
ratio of light to heavy chain mRNA levels as a predictor of aggregated
product levels, has been reported to help identify clones most
suitable for antibody production [17]. To monitor clone growth
characteristics, some level of screening is typically carried out in
shake flasks or even small bioreactors, in addition to early high-
throughput screening in 96-well plates [17]. In addition to material
proteins, DNA and high-molecular-weight aggregates to
acceptable levels in the drug substance [30]. To assure viral
clearance, the downstream process also includes at least
two dedicated, orthogonal steps for viral inactivation and
size-based viral removal by viral filtration. The last step in
the downstream process is typically an ultrafiltration/dia-
Trends in Biotechnology Vol.28 No.5
generation through the development of stable production cell lines,
the use of transient transfectant pools for production is frequently
carried out during pre-clinical developmental stages, to produce
rapidly significant quantities of the product [18,19], but is currently
rarely employed at scales beyond a few hundred liters of cell culture
volume. It is greatly preferred to lock in the most desirable subset of
clones for biopharmaceutical production early in the clinical devel-
opment process for a product. This avoids potential changes in
glycosylation profile later in development, which can create compar-
ability hurdles if a switch to an alternative cell line is undertaken.
In large-scale cell culture processes, the early stages of inoculum
growth after vial thaw are conducted in shake flasks or spinner flasks,
which progressively increase in size and/or volume. Wave bioreactors
in disposable bags are used extensively at this stage of the process.
The cell mass is then scaled up through several bioreactor stages
prior to transfer to the production bioreactor. Fed-batch production is
the most prevalent and involves the addition of small volumes of feed
to supplement the nutrients present in the bioreactor as cell growth
and product production progresses. Dissolved oxygen, pH, tempera-
ture and mass transfer of oxygen and CO2 are controlled in the
production bioreactor. Perfusion cell culture, in which the cells are
retained in the bioreactor and new media fed continuously is
employed in some processes, but is less widely adopted because of
challenges with maintaining sterility over long time periods. A key
development in the achievement of high cell culture titers has been an
improved understanding of nutrient limitations during cell culture
and the design of chemically defined, serum-free media that can
support cell growth and product secretion. Hydrolysates from yeast or
plant sources are often a component of cell culture media and have
effectively replaced the use of serum in large-scale processes. Amino
acids and trace media constituents such as metals have been shown
to influence product yields in cell culture [20,21]. Bioreactor processes
also have been developed to optimize gas exchange, to supply
sufficient oxygen to sustain cell growth and productivity, and to
remove CO2. Maintaining the efficiency of gas exchange is a very
important criterion for ensuring the success of scaling up cell culture
platforms [22].
mAbs are glycosylated products typically with a single N-linked site
on each half of the Fc region. Even though the N-linked glycosylation
heterogeneity does not significantly influence pharmacokinetics or
half-life of mAbs, because that is predominantly dictated by binding
to the neonatal Fc receptor (FcRn binding [23]), N-glycosylation does
play a role in mediating antibody-dependent cellular cytotoxicity or
complement-dependent cytotoxicity. Additionally, it is important to
prevent the occurrence of glycosylation structures that can trigger a
human immune response such as galactose a(1,3)-galactose and
NGNA structures [24]. As cell cultures under suboptimal conditions
have been shown to produce aberrant oligosaccharide structures
[20,21,25], it is important to link media and bioreactor condition
optimization to evaluation of glycosylation quality during cell culture
design.
A typical harvest procedure for mAb cell culture utilizes centrifuga-
tion followed by depth and membrane filters to remove cells and cell
debris prior to the purification steps in the downstream process [26].
Depth filters consist of a fibrous bed that can trap particles within their
volume in contrast to membrane filters that reject particulates.
Recently, depth filters have been shown to possess adsorptive
properties for soluble impurities such as host cell proteins and
DNA, in addition to their filtration role in removing particulates [27].
As a result, the type of depth filter and the choice of operating
conditions need to be studied in an integrative fashion with the
downstream purification process.
filtration (UF/DF) operation to reduce storage volumes and
exchange the product into the formulation buffer to pro-
duce the drug substance.
Alternatives to Protein A capture have been an area of
active investigation. As with any other protein, it is
possible to design a sequence of purification steps for mAbs
255
 Review
Figure 1. mAb upstream process. The process starts with vial thaw and expansion of cells through a series of inoculum steps. The cells are then further expanded in a series
of seed bioreactors before transfer to the production bioreactor where the mAb is expressed into the medium. Centrifugation and a series of filtration steps are then used to
harvest the cell culture broth from cells and cell debris.
without the use of an affinity step [31,32]. However, these
approaches are more difficult to template. In the past,
significant attention has focused on the development of
small-molecule ligands that could offer similar selectivity
to Protein A. However, comprehensive evaluation of hydro-
phobic charge induction chromatography (HCIC) [33] and
protein A mimetic ligands [34], using multiple antibody
process streams, has indicated markedly lower selectivity
than the Protein A ligand. Although the success of non-
proteinaceous ligands in replacing Protein A chromatog-
raphy has been low, these alternative modes of chromatog-
raphy are being employed as polishing steps in antibody
purification. Alternatively, other affinity ligands such as
single-chain antibodies, against the Fc region, derived from
camelid species have been shown to offer promising selec-
tivity for mAbs [35,36]. Improvements in the Protein A
ligand itself have also occurred, with one of the recent
improvements being the development of a base-stable ligand
that can resist alkaline conditions, which are often preferred
for sanitization of large-scale chromatography columns.
Box 2. Non-chromatographic technologies aimed at improving downstream process throughput
Renewed interest in non-chromatographic separation has been
sparked by some of the recent discussions about process throughput
limitations faced by batch chromatography. A number of such
techniques have been investigated ever since the advent of protein
production on a large scale, but have largely not been adopted for
therapeutic protein production. Aqueous two-phase separations
using polymer–salt [poly(ethylene glycol) (PEG) phosphate and PEG
citrate] systems have been applied successfully for purification of
mAbs from cell culture harvest streams [65]. Expanded bed chroma-
tography has been left behind during the past decade, despite
promise as an integrative technology between cell culture and
purification. Recently, interest has returned through the introduction
of improved resins and flow distribution systems that allow large-
scale implementation [66]. Precipitation of proteins using salts has
long been a mainstay of laboratory-level protein isolation. It has been
realized that application to mAb production will require improvement
in the selectivity of this technique [67–69]. Efforts focus on precipitat-
ing the product using negatively charged polyelectrolytes such as
256
Trends in Biotechnology Vol.28 No.5
This ligand also has improved the homogeneity of antibody
interactions with Protein A, thus facilitating the use of a
generic set of operating buffers for the Protein A step [37].
Recent advances in Protein A capture chromatography
Given the key significance of Protein A chromatography,
recent efforts have focused on improving its understanding
and mitigating some of the limitations it poses for large-scale
mAb production. Process throughput is constrained by the
maximum loading possible on Protein A; therefore, it is
important to understand why binding is limited. Using a
panel of mAbs and Fc fusion proteins, it has been determined
that multiple product molecules can bind to the Protein A
ligand. Inter- and intra-ligand steric effects have been shown
to limit the amount of mAb that can bind per unit volume of
the resin [38]. Dynamic binding capacity on the Protein A
affinity step is significantly influenced by flow rate and
residence time during loading [39]. The capture step, which
involves the loading of relatively dilute cell culture harvest
on the resin, is generally throughput-limited. To address this
caprylic acid or poly-ether sulfone. Some efforts have focused on
removing impurities by flocculation or precipitation. Promising
avenues for future research include the use of selective polymers
for product precipitation. Modified PEG molecules have been used for
selective capture of mAbs from CHO cell supernatants. Crystallization
can result in highly pure proteins and is often used for smaller
proteins and peptides. Although this area elicited some early interest
for mAb production, it is now realized that the phase behavior of
mAbs makes this a less robust process, with relatively narrow
operating ranges [70]. Charged UF membranes can provide separa-
tion based on size (hydrodynamic radius is strongly influenced by pH
and ionic strength) and charge [71,72]. Membrane adsorbers are
already a reality in the large-scale downstream processing of some
antibodies [73]. Anion exchange membrane adsorbers provide high
capacity and throughput but are effective for minimal polishing only
(DNA removal, some host cell protein clearance, and for providing
additional viral clearance capability). Ability to bind to large quantities
of aggregates or host cell protein is minimal.
 Review Trends in Biotechnology Vol.28 No.5

Figure 2. Downstream process platforms in use in several biopharmaceutical companies (a)–(c). The typical downstream platform starts with mAb capture by Protein A
chromatography. The commonly used platforms differ in the number and type of polishing steps. AEX, anion exchange; CEX, cation exchange; HIC, hydrophobic
interaction chromatography.

issue, when ample binding sites are available, a modulated
flow rate with rapid initial flow, followed by a progressive
decline in flow rate, has been developed [40].
Another area of concern for large-scale Protein A
chromatography is the lifetime of the resin, with re-use
over multiple cycles in operation. The Protein A ligand is
known to be sensitive to extreme conditions of pH and can
lose its ability to bind to products. The base-stable ligand
mentioned earlier is still not in widespread use because of
its higher cost. A progressive decline in binding capacity
has been shown to occur upon exposure to the mild alkaline
conditions often used for regenerating agarose-based
Protein A resins [41]. Protective additives in 0.1 N NaOH
[41] or use of a combination solution of phosphoric acid,
acetic acid and benzyl alcohol [42] have been found to be
effective sanitization solutions that maintain ligand func-
tionality. Overall, adsorbents based on Protein A have
been shown to be remarkably stable over many hundred
cycles of use, thus making Protein A chromatography a
very reliable tool in the antibody purification platform [43].
Host cell proteins constitute an important class of pro-
cess-related contaminants that need to be reduced to low
levels to assure product safety. While the Protein A capture
step is highly selective for mAbs over soluble host cell
protein impurities, the levels of host cell protein that often
persist after Protein A purification can still pose a clear-
ance challenge for non-affinity polishing steps. It has been
shown that host cell protein species that persist through

Figure 3. Monoclonal antibody downstream process. The process depicted in this figure starts with Protein A chromatographic capture and includes two subsequent
polishing chromatographic steps for impurity removal. The process also includes two dedicated orthogonal steps for viral clearance: low pH viral inactivation after Protein A
chromatography and viral filtration. The final process step is ultrafiltration/diafiltration (UF/DF) to formulate and concentrate the product.

257
Review
the Protein A step are often associated with the product
species [44]. These protein–protein interactions have been
found to be disrupted by chaotrope combinations at high
pH at which mAb–Protein A interactions are still strong.
This has allowed the development of selective wash con-
ditions that can reduce the host cell protein levels after
Protein A chromatography.
In addition to a better mechanistic understanding of
host cell protein clearance during mAb purification, better
methodologies have been developed to monitor host cell
protein (HCP) clearance and to validate the capability of
the downstream process to clear them. 2D difference gel
electrophoresis has been used to create a map of host cell
proteins to monitor their clearance through the down-
stream process [45]. Multiple strategies, including worst-
case studies, bypass and spiking, have been employed to
demonstrate robust clearance of host cell proteins through
a mAb downstream process [46]. The low-pH conditions
used for Protein A elution often result in precipitation
issues for the product species. These have been shown to
be mitigated by judicious buffer selection or by selecting
stabilizing additives for the elution buffer [47]. In some
cases, the turbidity has been shown to be caused by the
precipitation of host cell protein contaminants, rather than
the product. This can still be undesirable from a process
reproducibility and filtration perspective [48] and has been
shown to be prevented by appropriate sizing and design of
the harvest depth filtration step [27].
Recent advances in unit operations following Protein A
capture
As shown in Figure 2, some platforms adopt a completely
templated approach to the polishing steps (predominantly
through the use of anion (AEX) and cation (CEX) exchange
chromatography), whereas others have a molecule-specific
approach based on the particular impurity clearance chal-
lenge. Hydrophobic interaction chromatography (HIC) is
often selected for its particular ability to clear high-mol-
ecular-weight aggregates [6], and mixed-mode ion-
exchange chromatography has found recent application
as a result of its unique selectivity [49,50]. Several new
mixed-mode resins have been launched (e.g. Captoadhere1
and CaptoMMC1 from GE Healthcare) and are finding
increasing application as polishing steps for host cell
protein and aggregate clearance. Cation-exchange chroma-
tography column loading is often carried out at low salt
concentrations. mAbs have a basic pI, and the initial
molecules that adsorb on the resin are often at the entrance
to pores in the resin structure, thus repelling more protein
molecules and restricting column load capacities [51]. The
anion-exchange column is usually operated under flow-
through conditions at fairly high load capacities. As
described later, this step can sometimes be replaced with
a membrane adsorber step. If impurity resolution, particu-
larly HCP and aggregate is needed, the anion-exchange
chromatographic step can be operated under a weak par-
titioning mode as opposed to true flow-through conditions,
which results in retention of the impurities [52].
A majority of recently developed mAb production pro-
cesses employ parvoviral grade viral filters (20 nm reten-
tion size) in contrast to the retroviral grade filters (50 nm
258
Trends in Biotechnology Vol.28 No.5
retention size) that were employed in the past [53].
Although the smaller pore size rating provides added viral
retention, these filters can be clogged by small levels of
particulates that are otherwise not detectable by analytical
size exclusion chromatography (SEC) or particle size
measurements in the product stream [54,55]. Viral filters
are typically single-use filters and can be among the most
expensive consumables in the process based on their cost
per gram of product. In addition to appropriate placement
of these filters in the downstream process, pre-filters can be
used to protect the viral filters from flux decay arising from
a high particulate load [56]. Viral filters from multiple
vendors are typically screened to arrive at the right selec-
tion for a process, and significant innovation has taken
place in recent years in the development of low-fouling
viral filters that can still provide excellent viral clearance
[57]. These filters are made of hydrophilic materials that
reduce the tendency for proteins to unfold at the surface,
thus minimizing nonspecific fouling of the pores.
Additional improvements have been made in pore
morphology, with a reduction in polydispersity of pore
sizes. These have led to progressive improvement in the
maintenance of high fluxes with mAb process streams.
The large production quantities of mAbs create another
challenge for the handling of the large volumes of drug
substance produced. Several ultra-high-concentration
antibody formulations with protein concentrations exceed-
ing 100 mg/ml are in use for mAbs, particularly for sub-
cutaneous delivery. The generation of such high protein
concentrations requires special attention to the design of
the UF/DF step that is typically used for buffer exchange
into the formulation buffer [58]. Air–water interfaces cre-
ated during pumping of such highly concentrated protein
solutions can cause product aggregation during large-scale
production [59]. The flowpath in the UF system is designed
to minimize system hold-up volumes and to decrease the
incidence of zones where air entrainment might occur. For
high-concentration formulations, the concentration at
which DF is conducted is selected to minimize the number
of pump passes to which an average protein molecule
might be exposed during the step. Even exposure to stain-
less steel surfaces for significant lengths of time can result
in aggregation [60]. If the drug substance requires frozen
conditions during long-term storage, design of the freezing
and thawing steps needs to be carefully approached to
avoid product quality changes caused by cryo-concen-
tration and exclusion effects caused by slow freezing rates
in large volumes, and product denaturation on high-sur-
face-area ice crystals caused by rapid freezing rates [61].
The adoption of a platform approach has created greater
efficiencies in process development and manufacturing for
mAbs. Evolution of the platform through experiences accu-
mulated while dealing with challenges faced with atypical
antibody processes is an equally essential part of creating a
robust and widely applicable platform.
Non-chromatographic downstream processing
Chromatographic steps have formed the mainstay of mod-
ern-day biopharmaceutical purification processes because of
their high selectivity and resolution. However, throughput
for conventional preparative chromatography is inherently
Review
limited as a result of the batch nature of the process and the
limitations in fluid velocity for medium-sized particle beads
(40–120 mm) that are used in these columns. With increas-
ing production amounts per cell culture batch in commercial
scale operations (up to 25,000 l working volumes), the
achievement of >50 kg drug substance batches per run
are close to realization. Significant debate has been sparked
by these changes over whether conventional packed bed
chromatography can continue to meet downstream process
demands in the future. Persuasive arguments have been
made for two-column step processes with high product loads
on Protein A capture, and a single polishing step operated in
the flow-through mode, which indicate that chromato-
graphic steps with adequate cycling can continue to meet
these demands into the future [7,8]. It has also been argued
that very few mAb products actually require these large
production scales because of limitations on market demand.
Others have argued that innovative unit operations are
worthwhile investigating as alternative means of dealing
with ever-increasing bioreactor output [29,62,63]. Unit
operations such as aqueous two-phase separation, selective
precipitation with polyelectrolytes or polymers, membrane
chromatography, and high-resolution UF and crystalliza-
tion have been cited as possible replacements for chroma-
tography.
In reality, the best non-chromatographic processing
steps will likely resemble a combination of these two
seemingly divergent scenarios. It is clear that the biophar-
maceutical industry is slow to adopt novel processing
technologies that require significant capital investments
or modifications to existing facilities. Globally, biopharma-
ceutical manufacturers have made significant investments
in current processing equipment, particularly centrifuges
and large-scale tanks. By contrast, the industry has readily
adopted innovations that fit in within the current facility
paradigm, such as novel adsorbents and membrane
chromatography.
Understanding some of the reasons behind cost press-
ures is key in predicting what changes could occur in the
future. A pressure that has been cited as a reason for
significant change is the coming advent of follow-on bio-
logics [64]. However, follow-on biologics (also known as
‘biosimilars’) will be required to face regulatory require-
ments for comparability, which is a significantly bigger
hurdle for biopharmaceuticals than for small-molecule
pharmaceuticals. The burden in demonstrating compar-
ability for these complex molecules often will extend
beyond biochemical and biophysical comparability to in-
clude safety and efficacy clinical trials. This is likely to
diminish the potential for biosimilars to gain a significant
foothold in the marketplace. As a result, we feel it is
unlikely that the cost pressures from biosimilars will serve
as a major driver for innovation in biologic manufacturing
processes. By contrast, biopharmaceutical companies
already face significant internal pressures to minimize
extensive investments in new facilities and optimize the
usage of existing facilities. A large part of this pressure is
due to the slowing rate of new drug approvals and
increased competition among pharmaceutical and biotech-
nology companies in indications for which new thera-
peutics are developed. Effective utilization of existing
Trends in Biotechnology Vol.28 No.5
manufacturing facilities then becomes the major driver
for process innovation.
Innovative technologies that can operate within the
current plant framework are likely to be the first movers
in a changing processing landscape. Box 2 lists some of the
non-chromatographic technologies that are being investi-
gated to improve process throughput for mAb downstream
processing.
Production of antibody-related products
In addition to intact mAbs, of which >20 have been
approved for therapeutic use, there is growing interest
in antibody-related proteins as the next wave of biother-
apeutics. These include Fc fusion proteins, which consist of
a fusion between the Fc region of mAbs and a protein
partner that is often the ligand for a target protein [74].
Four Fc fusion proteins are currently approved for thera-
peutic use. Chemical conjugates with antibodies are
advancing in clinical trials, particularly in oncology [75].
Three conjugated antibodies are commercially approved
(Zevalin, Bexxar and Mylotarg). Although they possess
several advantages, intact mAbs often have limited bioa-
vailability because of their large size. Antibody fragments
are being developed to circumvent this limitation, while
retaining the binding characteristics of mAbs [76]. A new
generation of products that utilize scaffolds, and are
similar to mAbs, have also been developed. These include
adnectins that are based on the fibronectin III protein and
can be engineered to bind to a wide range of targets in
similar fashion to mAbs [77,78]. In addition to improved
bioavailability to tissues, these engineered proteins can
potentially access clefts or grooves within target antigens
that bigger mAbs may not.
The mAb process platform can serve the needs for some
of these types of products, including Fc fusion proteins,
bispecific antibodies or conjugated antibodies. Antibody
fragments are smaller units that are not glycosylated,
which allows high productivity that is inherent in bacterial
expression systems [79]. Many of these types of products
are expressed as intracellular inclusion bodies that require
refolding to recover the intact product. Refolding of the
inclusion bodies is carried out using chaotropes, such as
urea, at low protein concentrations to reduce product
aggregation as the denaturant concentration is decreased.
The polishing steps for these products need to achieve
separation of the misfolded forms in addition to separ-
ated-out host cell proteins and aggregates. Dedicated viral
clearance steps are not a regulatory requirement for micro-
bially expressed proteins. Refolding can often be a low-
yielding unit operation, and, as mammalian cell culture
titers improve, there could be occasion to revisit the expres-
sion system of choice for antibody fragments. A defined
process platform has yet to become widely established for
antibody fragment production but will be a key future
development as these products mature in the development
pipeline.
Challenges and future trends
The upstream process platform for mAb production has
evolved significantly, both on the cell line creation and
production bioreactor process fronts, which leads to the
259
Review
achievement of product titers that have exceeded all
previous experience with cell culture. The downstream
process platform for mAb production still includes several
chromatographic steps. Improved understanding of pro-
cess-scale purification of mAbs has arisen as a result of
experience with an ever-increasing number of mAbs in
development and large-scale manufacturing. Future devel-
opments will take place on two fronts: technology that will
enable incremental advances in the current platforms, and
developments that will cause more radical change in the
technology employed for mAb production.
Although the current process platform might appear to
be safely ensconced in practice, we feel that several
drivers will lead to significant evolution over the next
decade. With the high titers now achievable in cell culture,
the need to construct new facilities with very large
bioreactors will diminish, although current facilities will
continue to be in use. Lower production volumes will
increase the need for facility flexibility and faster
turnaround that will lead to growth in the use of
disposables in manufacturing. On the process front, the
productivity driver will remain even with the use of
smaller bioreactors. This will continue to drive innovation
in the development of non-chromatographic downstream
steps. However, it has to be noted that, while such
technologies have been investigated for a very long time
already, implementation into a biopharmaceutical
process platform has not yet been demonstrated. We feel
that the alternative technologies that will be the first to
find application will be those that fit within the current
biomanufacturing facility designs, rather than requiring
major investment on the facilities front.
Finally, another key development will be the matu-
ration of antibody fragments and novel scaffolds within
the clinical development pipelines. We anticipate the birth
of a new generation of platform processes that address
specific manufacturing issues with these therapeutic mod-
alities, using a combination of traditional and non-
traditional methods.
References
1 Reichert, J. and Valge-Archer, V. (2007) Development trends for
monoclonal antibody cancer therapeutics. Nat. Rev. Drug Discov. 6,
349–356
2 Carter, P. (2006) Potent antibody therapeutics by design. Nat. Rev.
Immunol. 6, 343–357
3 Ruuls, S. et al. (2008) Novel human antibody therapeutics: the age of
the Umabs. Biotechnol. J. 3, 1157–1171
4 Jefferis, R. (2009) Glycosylation as a strategy to improve antibody-
based therapeutics. Nature 8, 226–234
5 Weiner, L. (2006) Fully human therapeutic monoclonal antibodies J.
Immunotherapy 29, 1–9
6 Shukla, A. et al. (2007) Downstream processing of monoclonal
antibodies – application of platform approaches. J. Chromatogr. B
848, 28–39
7 Kelley, B. (2007) Very large-scale monoclonal antibody purification –
the case for conventional unit operations. Biotechnol. Prog. 23, 995–
1008
8 Kelley, B. (2009) Industrialization of mAb production technology – the
bioprocessing industry at a crossroads. MAbs 1, 1–10
9 Birch, J. and Racher, A. (2006) Antibody production. Adv. Drug Deliv.
Rev. 58, 671–685
10 Chusainow, J. et al. (2009) A study of monoclonal antibody producing
CHO cell lines – what makes a stable high producer? Biotechnol.
Bioeng. 102, 1182–1196
260
Trends in Biotechnology Vol.28 No.5
11 Jiang, Z. et al. (2006) Regulation of recombinant monoclonal
antibody production in CHO cells – a comparative study of gene
copy number, mRNA level and protein expression. Biotechnol. Progr.
22, 313–318
12 Wurm, F. (2004) Production of recombinant protein therapeutics in
cultivated mammalian cells. Nat. Biotechnol. 22, 1393–1398
13 Chu, L. and Robinson, D. (2001) Industrial choices for protein
production by large-scale cell culture. Curr. Opin. Biotechnol. 12,
180–187
14 Wagner-Rousset, E. et al. (2008) The way forward, enhanced
characterization of therapeutic antibody glycosylation: comparison of
three level MS based strategies J. Chromatogr. B 872, 23–37
15 Majors, B. et al. (2009) Mcl-1 overexpression leads to higher viabilities
and increased production of humanized monoclonal antibody in CHO
cells. Biotechnol. Progr. 25, 1161–1168
16 Jain, E. and Kumar, A. (2008) Upstream processes in antibody
production – evaluation of critical parameters. Biotechnol. Adv. 26,
46–72
17 Lee, C. et al. (2008) A clone screening method using mRNA levels to
determine specific productivity and product quality for monoclonal
antibodies. Biotechnol. Bioeng. 102, 1107–1118
18 Haldankar, R. et al. (2006) Serum free suspension large-scale transient
tranfection of CHO cells in WAVE bioreactors. Mol. Biotechnol. 34,
191–199
19 Ye, J. et al. (2009) High-level protein expression in scaleable CHO
transient transfection. Biotechnol. Bioeng. 103, 542–551
20 Crowell, C. et al. (2007) Amino acid and manganese supplementation
modulates the glycosylation state of erythropoietin in a CHO culture
system. Biotechnol. Bioeng. 96, 538–549
21 Wong, D. et al. (2004) Impact of dynamic on-line fed-batch strategies on
metabolism, productivity and N-glycosylation in CHO cell cultures.
Biotechnol. Bioeng. 89, 164–177
22 Matsunaga, N. et al. (2009) Estimation of dissolved carbon dioxide
stripping in a large bioreactor using model medium. J. Biosci. Bioeng.
107, 49–424
23 Roopenian, D. and Akilesh, S. (2007) FcRn: the neonatal Fc receptor
comes of age. Nat. Rev. Immunol. 7, 715–725
24 Walsh, G. and Jefferis, R. (2006) Post-translational modifications
in the context of therapeutic proteins. Nat. Biotechnol. 24, 1241–
1252
25 Hossler, P. et al. (2009) Optimal and consistent protein glycosylation in
mammalian cell culture. Glycobiology 19, 936–949
26 Shukla, A. and Kandula, J. (2008) Harvest and recovery of monoclonal
antibodies from large-scale mammalian cell culture. Biopharm. Int. 7,
34–45
27 Yigzaw, Y. et al. (2006) Exploitation of the adsorptive properties of
depth filters for host cell protein removal during monoclonal antibody
purification. Biotechnol. Progr. 22, 288–296
28 Kozlowski, S. and Swann, P. (2006) Current and future issues in the
manufacturing and development of monoclonal antibodies. Adv. Drug
Deliv. Rev. 58, 707–722
29 Low, D. et al. (2007) Future of antibody purification. J. Chromatogr. B
848, 48–63
30 Ghose, S. et al. (2009) Integrated polishing steps for monoclonal
antibody production, In Process Scale Purification of Antibodies (1st
edn) (Gottschalk, U., ed.), pp. 145–167, Wiley
31 Follman, D. and Fahrner, R. (2004) J. Chromatogr. A 1024, 79–85
32 Arunakumari, A. and Wang, J. (2009) Purification of human
monoclonal antibodies – non Protein A strategies, In Process Scale
Purification of Antibodies (1st edn) (Gottschalk, U., ed.), pp. 103–123,
Wiley
33 Ghose, S. et al. (2005) Protein interactions in hydrophobic
charge induction chromatography (HCIC). Biotechnol. Progr. 21,
498–508
34 Ghose, S. et al. (2006) Evaluation and comparison of alternatives to
Protein A chromatography – mimetic and hydrophobic charge
induction chromatographic stationary phases. J. Chromatogr. A
1122, 144–152
35 Liu, J. et al. (2009) Comparison of camelid antibody ligand to Protein A
for monoclonal antibody purification. Biopharm. Int. 22, 35–43
36 Zandian, M. and Jungbauer, A. (2009) Engineering properties of a
camelid antibody affinity sorbent for immunoglobulin G purification. J.
Chromatogr. A 1216, 5548–5556
Review
37 Ghose, S. et al. (2005) Antibody variable region interactions with
Protein A – implications for the development of generic purification
processes. Biotechnol. Bioeng. 92, 665–673
38 Ghose, S. et al. (2007) Binding capacity differences for antibodies and
Fc-fusion proteins on Protein A chromatographic materials.
Biotechnol. Bioeng. 96, 768–779
39 Tugcu, N. et al. (2008) Maximizing productivity of chromatographic
steps for purification of monoclonal antibodies. Biotechnol. Bioeng. 99,
599–613
40 Ghose, S. et al. (2004) Use and optimization of a dual flow-rate loading
strategy for maximizing throughput in Protein A affinity
chromatography. Biotechnol. Progr. 20, 830–840
41 Jiang, C. et al. (2009) A mechanistic study of Protein A chromatography
resin lifetime. J. Chromatogr. A 1216, 5849–5855
42 Rogers, M. et al. (2009) Development of a rapid sanitization solution for
silica-based protein A affinity adsorbents. J. Chromatogr. A 1216,
4589–4596
43 Brorson, K. et al. (2003) Identification of protein A media performance
attributes that can be monitored as surrogates for retrovirus clearance
during extended re-use. J. Chromatogr. A 989, 155–163
44 Shukla, A. and Hinckley, P. (2008) Host cell protein clearance during
Protein A chromatography – development of an improved wash step.
Biotechnol. Progr. 24, 1115–1121
45 Jin, M. et al. (2010) Profiling of host cell proteins by two dimensional
difference gel electrophoresis (2D-DIGE) – implications for
downstream process development. Biotechnol. Bioeng. 105, 306–316
46 Shukla, A. et al. (2008) Demonstration of robust host cell protein
clearance in biopharmaceutical downstream processes. Biotechnol.
Progr. 24, 615–622
47 Shukla, A. et al. (2007) Protein aggregation kinetics during Protein A
chromatography – case study for an Fc fusion protein. J. Chromatogr. A
1171, 22–28
48 Kandula, S. et al. (2009) Design of a filter train for precipitate removal
in mAb downstream processing. Biotechnol. Appl. Biochem. 54, 149–
155
49 Gagnon, P. (2009) IgG aggregate removal by charged hydrophobic
mixed-mode chromatography. Curr. Pharm. Biotech. 10, 434–439
50 Zhao, G. et al. (2009) Ligands for mixed-mode protein chromatography:
principles, characteristics and design. J. Biotechnol. 144, 3–11
51 Zydney, A. et al. (2009) Modeling electrostatic exclusion effects during
ion exchange chromatography of monoclonal antibodies. Biotechnol.
Bioeng. 102, 1131–1140
52 Kelley, B. et al. (2008) Weak partitioning chromatography for anion-
exchange purification of monoclonal antibodies. Biotechnol. Bioeng.
101, 553–566
53 Zhou, J. (2008) Viral clearance using disposable systems in monoclonal
antibody commercial downstream processing. Biotechnol. Bioeng. 100,
488–496
54 Kern, G. and Krishnan, M. (2006) Virus removal by filtration – points to
consider. Biopharm. Int. 19, 32–41
55 Bohonak, D. and Zydney, A. (2005) Compaction and permeability
effects with virus filtration membranes. J. Membrane Sci. 254, 71–79
56 Ireland, T. et al. (2005) Optimizing virus filtration performance with
prefiltration. Bioprocess Intl. Supplement 3, 44–47
57 Marques, B. et al. (2009) Virus filtration of high-concentration
monoclonal antibody solutions. Biotechnol. Progr. 25, 483–491
Trends in Biotechnology Vol.28 No.5
58 Harris, R. et al. (2004) Commercial manufacturing scale formulation
and analytical characterization of therapeutic recombinant antibodies.
Drug Dev. Res. 61, 137–154
59 Bee, J. et al. (2009) Response of a concentrated monoclonal antibody
formulation to high shear. Biotechnol. Bioeng. 103, 936–943
60 Bee, J. et al. (2010) Aggregation of a monoclonal antibody induced by
adsorption to stainless steel. Biotechnol. Bioeng. 105, 121–129
61 Singh, S. et al. (2009) Large-scale freezing of biologics. Bioprocess Intl.
7, 32–44
62 Przybycien, T. et al. (2004) Alternative bioseparation operations – life
beyond packed-bed chromatography. Curr. Opin. Biotech. 15, 469–478
63 Thommes, J. and Etzel, M. (2007) Alternatives to chromatographic
separations. Biotechnol. Progr. 23, 42–45
64 Grabowski, H. (2008) Follow-on biologics – data exclusivity and the
balance between innovation and competition. Nat. Rev. Drug. Discov. 7,
478–488
65 Azevedo, A. et al. (2009) Chromatography-free recovery of
biopharmaceuticals through aqueous two-phase processing. Trends
Biotechnol. 27, 240–247
66 Noel, R. (2007) Chromatography: industrial processing and
biotherapeutics Filtration and Separation 44, 26–28
67 Wang, J. et al. (2009) Precipitation of process-derived impurities in
non-Protein A purification schemes for antibodies. Biopharm Intl.
Special Issue on Downstream Processing 2010 – embracing
innovation Oct, 4–10
68 McDonald, P. et al. (2009) Selective antibody precipitation using
polyelectrolytes – a novel approach to the purification of monoclonal
antibodies. Biotechnol. Bioeng. 102, 1141–1151
69 Azevedo, A. et al. (2009) Affinity enhanced purification of human
monoclonal antibodies by aqueous two-phase separation. Sepn.
Purific. Technol. 65, 31–39
70 Ahamed, T. et al. (2007) Phase behavior of an intact monoclonal
antibody. Biophys. J. 93, 610–619
71 van Reis, R. and Zydney, A. (2007) Bioprocess membrane technology. J.
Membrane Sci. 302, 16–50
72 Rao, S. and Zydney, A. (2006) High resolution protein separations
using affinity ultrafiltration with small charged ligands. J. Membrane
Sci. 280, 781–789
73 Zhou, J. and Tressel, T. (2006) Basic concepts in Q membrane
chromatography for large-scale antibody production. Biotechnol.
Progr. 22, 341–349
74 Chang, C. et al. (2009) Advances and challenges in developing cytokine
fusion proteins as improved therapeutics. Expert Opin. Drug. Discov. 4,
181–194
75 Deonarain, M. (2008) Recombinant antibodies for cancer therapy.
Expert Opin. Biol. Ther. 8, 1123–1141
76 Holliger, P. and Hudson, P. (2005) Engineered antibody fragments and
the rise of single domains. Nat. Biotechnol. 23, 1126–1136
77 Gill, D. and Damle, N. (2006) Biopharmaceutical drug discovery using
novel protein scaffolds. Curr. Opin. Biotechnol. 17, 653–658
78 Dineen, S. et al. (2008) The adnectin CT-322 is a novel VEGF receptor
inhibitor that decreases tumor burden in an orthotopic mouse model of
pancreatic cancer. BMC Cancer 8 (352), 1–10
79 Andersen, D. and Reilly, D. (2004) Production technologies for
monoclonal antibodies and their fragments. Curr. Opin. Biotechnol.
15, 456–462
261