15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Performance Optimization of Continuous Countercurrent Tangential
Chromatography for Antibody Capture
Amit K. Dutta, Jasmine Tan, and Boris Napadensky
Chromatan Corporation, 200 Innovation Blvd, Suite 260B, State College, PA 16803

Andrew L. Zydney
Dept. of Chemical Engineering, The Pennsylvania State University, University Park, PA 16802
Oleg Shinkazh
Chromatan Corporation, 200 Innovation Blvd, Suite 260B, State College, PA 16803

DOI 10.1002/btpr.2250
Published online March 17, 2016 in Wiley Online Library (wileyonlinelibrary.com)

Recent studies have demonstrated that continuous countercurrent tangential chromatography

(CCTC) can effectively purify monoclonal antibodies from clarified cell culture fluid. CCTC
has the potential to overcome many of the limitations of conventional packed bed protein A
chromatography. This paper explores the optimization of CCTC in terms of product yield,
impurity removal, overall productivity, and buffer usage. Modeling was based on data from
bench-scale process development and CCTC experiments for protein A capture of two clarified
Chinese Hamster Ovary cell culture feedstocks containing monoclonal antibodies provided by
industrial partners. The impact of resin binding capacity and kinetics, as well as staging strat-
egy and buffer recycling, was assessed. It was found that optimal staging in the binding step
provides better yield and increases overall system productivity by 8–16%. Utilization of higher
number of stages in the wash and elution steps can lead to significant decreases in buffer
usage (40% reduction) as well as increased removal of impurities (2 log greater removal).
Further reductions in buffer usage can be obtained by recycling of buffer in the wash and
regeneration steps (35%). Preliminary results with smaller particle size resins show that the
productivity of the CCTC system can be increased by 2.5-fold up to 190 g of mAb/L of resin/
hr due to the reduction in mass transfer limitations in the binding step. These results provide
a solid framework for designing and optimizing CCTC technology for capture applications.
C 2016 American Institute of Chemical Engineers Biotechnol. Prog., 32:430–439, 2016
V
Keywords: continuous chromatography, monoclonal antibody, protein A, optimization, anti-
body purification

Introduction the binding, wash, elution, and regeneration steps occurring

Monoclonal antibody (mAb) based treatments of cancer
and autoimmune diseases have been established as one of
the most successful therapeutic strategies. This class of drugs
R R
includes such blockbusters as HerceptinV, RemicadeV,
R R
HumiraV, RituxanV, and many others. There are currently
over 300 mAbs in various stages of clinical trials. These
products have driven much of the growth in the biotechnol-
ogy industry over the last 20 years.1,2
Protein A column chromatography is currently used for ini-
tial capture of nearly all mAb products from clarified cell cul-
ture fluid (CCF). Protein A provides a highly robust method
for use in platform processes, and it is currently considered
the most practical capture methodology for commercial pro-
duction of monoclonal antibodies.3 Protein A capture systems
typically use large stainless steel columns in batch mode with
Additional Supporting Information may be found in the online ver-
sion of this article.
Correspondence concerning this article should be addressed to O.
Shinkazh at oleg.shinkazh@chromatan.com.
sequentially (one after another). While effective and robust,
this method of manufacturing has caused major bottlenecks in
downstream processing.4 The problem has become especially
severe because of increasing bioreactor titers over the past 10
years in combination with the increased demands on cost
reduction in mAb manufacturing.5 Additional drawbacks of
batch chromatography include low productivity (5–15 g of
mAb/L of resin/hr is typical), high pressure operation, nonlin-
ear scalability for large scale operations, and inability to inter-
face with continuous production systems such as perfusion
bioreactors. Because of these inefficiencies, affinity columns
use very large volumes of resin that cost $10,000–$17,000 per
liter. A single large scale protein A column designed for pri-
mary mAb capture can therefore cost manufacturers $15 mil-
lion for the resin alone.6
There is considerable interest throughout the biopharma-
ceutical industry in moving from batch to continuous proc-
esses. Continuous processing is currently the standard mode
of operation in the chemical and food industries. There are
several advantages of continuous processing including: (1)

430 C 2016 American Institute of Chemical Engineers

V

Biotechnol. Prog., 2016, Vol. 32, No. 2
better quality control, (2) much lower capital cost, (3) lower
equipment footprint, and (4) enhanced flexibility in manufac-
turing.7,8 Another trend in bioprocessing is the application of
single-use equipment which can remove some of the major
hurdles associated with biopharma manufacturing, such as
cleaning, column packing, and validation while reducing the
overall cost of production.
Several efforts have been reported to develop continuous
bioprocessing platforms. For example, Warikoo et al.9 and
Godawat et al.10 used periodic counter-current (PCC) multi-
column chromatography for continuous capture of mAb from
a perfusion bioreactor. They reported approximately five times
higher volumetric productivity than a fed batch process with
comparable yield and purity while the PCC column size was
smaller than that used in the fed-batch process.9 In another
study, Pollock et al.11 studied the design and optimization of
a PCC system. Muller-Spath et al.12 have used multicolumn
countercurrent solvent gradient purification (MCSGP) chroma-
tography with anion and cation exchange resins for antibody
purification. One problem with multicolumn chromatography
is that these systems require complex valve switching and
packing of multiple columns (unless operated with more
expensive prepacked columns). Also, they operate in a cyclic
rather than true steady state mode, which can lead to greater
variability in product quality while creating challenges for
subsequent downstream operations.13 A number of companies
have recently commercialized prepacked columns that could
be applied in single-use applications. However, in manufactur-
ing, these are mostly targeted for multibatch clinical cam-
paigns in order to maximize utilization of the expensive
resins, particularly protein A, to their full life cycles. Rosa
et al.14 have examined aqueous two-phase extraction while
Hammerschmidt et al.15 have explored the use of precipitation
as alternatives to chromatography for steady state capture of
product. However, unlike protein A affinity resins, these sys-
tems typically cannot provide robust operation and high
purity. Note that the membrane adsorbers developed by Natrix
Separations Inc.16 are single-use, but they currently can only
perform batch separations.
The continuous countercurrent tangential chromatography
(CCTC) system developed by Chromatan Corporation con-
sists of a series of chromatographic steps, with each step
consisting of several stages. A CCTC stage is defined as a
combination of a single hollow fiber membrane module and
a static mixer.17 A schematic of the system is shown in Fig-
ure 1A and includes six steps: binding, wash-1, wash-2, elu-
tion, stripping, and equilibration. The inset shows
countercurrent staging employed in the elution step. Similar
designs are used for the other steps (discussed in more detail
subsequently). The static mixer provides contacting between
the resin slurry and buffer/cell culture fluid (CCF) with very
uniform residence time distribution. The outlet from the
static mixer is connected to a hollow fiber membrane module
(shown schematically in Figure 1B). The pore size of the
membrane (0.5 mm) is much smaller than that of the resin
(10–50 mm), but much larger than the mAb and any molecu-
lar impurities (5–20 nm). The resin slurry flows through the
lumen (retentate) side while the product (or impurities) are
removed in the permeate that is collected through the mem-
brane. In the binding step, fresh resin is mixed with CCF
which allows for capture of the mAb. The resin slurry with
bound mAb is then passed to the washing steps where the
unbound impurities are removed as waste in the permeate
while the resin with bound mAb is passed to the next step
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
431
via the retentate exit (the lumen side outflow). A similar
design is used in the subsequent steps.
The number of stages in the binding, washing, elution,
stripping, and regeneration steps can be adjusted to optimize
system performance—this is discussed in more detail in the
body of this paper. A two stage binding step is shown in Fig-
ure 1C. In this case, the fresh resin is introduced via the first
stage and is combined with partially bound mAb solution that
is recycled from the second stage. Fresh CCF is loaded on to
the second stage and combined with partially loaded resin
from the first stage. This creates a unique countercurrent bind-
ing configuration that is unavailable in conventional column
chromatography. The resin from the second stage continues to
the “after binder” which binds all residual mAb before enter-
ing the wash-1 step, while the CCF waste leaves the system
as first stage permeate. The impact of the after binder on
product yield is discussed subsequently.
CCTC overcomes many of the limitations of conventional
packed bed chromatography. The system operates at low
pressure using a fully disposable flow path, while the product
is loaded and purified in a truly continuous steady state fash-
ion. This is a major advantage vs. multicolumn chromatogra-
phy systems which operate in a continuous mode, but have a
cyclic product discharge, causing eluted product concentra-
tions to vary widely. While the CCTC system discharges a
relatively dilute mAb concentration in comparison with that
obtained from a multicolumn system, an in-line ultrafiltration
system can be used in a straightforward manner to control
the final product pool concentration to any desired level.
Another key advantage of the CCTC system is that the pres-
sure drop is independent of the particle size, in sharp con-
trast to the behavior of packed columns. Thus, small particle
size resins (10 mm) with low crosslinking densities can be
used in the CCTC system; these small particles will have
much faster mass transfer rates, leading to greater overall
productivity.
Shinkazh et al.17 used CCTC for the separation of bovine
serum albumin (BSA) and myoglobin using an anion
exchange resin. CCTC has also been used for purification of
an IgG4 from a mixture of BSA and myoglobin using a
protein A affinity resin.18 More recently, Dutta et al.19 dem-
onstrated the ability of CCTC to purify monoclonal antibod-
ies produced in CHO cells. Two different mAbs, with low
and high titers, were purified directly from clarified cell cul-
ture fluid using the CCTC system, with the overall produc-
tivity nearly five times the productivity of a packed column
with similar yield and purity.
The objective of this study was to examine the design and
optimization of the CCTC system focusing on the mAb
yield, the system productivity, and buffer utilization. Model
calculations were used to examined the effects of staging,
particle size, and resin capacity on the performance of the
CCTC system.
Model
The CCTC system consists of a series of static mixers and
hollow fiber membrane modules. A diagram showing all the
streams for a single static mixer and membrane are shown in
Figure 2. The governing mass balances for each static mixer
can be written as:
qs;in 1 qw 5 qs;out (1)

432
Figure 1. (A) Schematic of the CCTC system. Each step consists of a combination of static mixers and hollow fiber membrane mod-
ules (see inset for the elution step). The pumps are not shown for simplicity. (B) Schematic of the hollow fiber module. (C)
Schematic of a two stage binding step. BP and BR represent the binding step permeate and retentate streams, respectively.19
qs;in uin 5 qs;out uout
qs;in uin Cb;in 1 qs;in ð12uin ÞCf;in 1 qw Cw 5
qs;out uout Cb;out 1 qs;out ð12uout ÞCf;out
where qs,in and qs,out are the slurry volumetric flow rates
in the inlet and outlet streams; qw is the volumetric fluid
flow rate into the mixer (CCF, wash buffer, elution buffer,
etc.); uin and uout are the resin volume fractions in the
inlet and outlet streams; Cb,in and Cb,out are the concentra-
tions of bound antibody in the inlet and outlet streams;
and Cf,in, Cf,out, and Cw are the concentrations of free anti-
body in the inlet, outlet, and fluid streams. The corre-
sponding mass balances for the hollow fiber membrane
modules are:
qs;out 5 qr 1 qp
qs;out uout 5 qr ur
qs;out uout Cb;out 1 qs;out ð12uout ÞCf;out
5 qr ur Cb;r 1 qr ð12ur ÞCf;r 1 qp Cp
where the subscripts r and p refer to the retentate and perme-
ate streams. Since the membrane is assumed to be fully per-
meable to the antibody, host cell proteins, and other
impurities:
Cp 5 Cf;r
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biotechnol. Prog., 2016, Vol. 32, No. 2
(2) Binding step
The binding step provides contacting between the clean
(3) resin and the clarified cell culture fluid containing the mAb.
Thus, in a system with only a single binding stage, Cb,in 5
Cf,in 5 0 and Cw 5 CF is the mAb concentration in the CCF
feed. Cb,out and Cf,out are determined by the binding kinetics
based on the residence time in the static mixer as described
in Supporting Information. Additional details on the kinetics
model are available in our previous publication.19 The resi-
dence time in the membrane was much smaller than that in
the static mixer and was neglected in the binding calcula-
tions; thus, Cb,r 5 Cb,out, and Cf,r 5 Cf,out 5 Cp. The mAb
loading (L) is given as:
L5 Cb;out =Qmax 5 cb CF =ðQmax uin Þ (8)
(4)
cb 5qw =qs;in (8A)
(5)
where cb is the ratio of the CCF to resin flow rates which is
one of the key design variables.
(6)
The productivity of the binding step can be increased by
using a two stage (countercurrent) binding step, which pro-
vided equivalent mAb capture with less total residence time
(i.e., static mixer volume); this is discussed in more detail in
the “Results” section. Calculations were performed using the
same basic approach as for single stage binding, but in this
case, the CCF was fed to the second stage and mixed with
(7) resin that was preloaded with a small amount of mAb from
the first stage. The permeate from the second stage is then

Biotechnol. Prog., 2016, Vol. 32, No. 2
Figure 2. Schematic of (A) static mixer and (B) membrane module showing all inlet and outlet streams.
recycled to the first stage, where most of the remaining
unbound mAb is bound by the fresh slurry. Permeate from
the first stage then leaves as waste. In addition, the two stage
system included an after binder, a static mixer placed after
the final hollow fiber membrane module, to capture the small
amount of residual mAb remaining in free solution in the
final retentate stream as shown in Figure 1C.
The binding step capture yield (Ybind) for a single stage
system was evaluated from the following equation:
 
Cf ;r
Ybind 5 12 (9)
C0
since the slurry flow rate out of the hollow fiber module was
controlled so that it is equal to the slurry flow rate entering
the first static mixer, i.e., qr 5 qs,in. C0 is the mAb concentra-
tion in the feed immediately after mixing between CCF and
resin slurry (before any binding occurs):
C q
C0 5   F w  (9A)
qw 1 qs;in ð12uin Þ
Wash step
The wash step is designed to remove unbound impurities
(e.g., host cell proteins and DNA) in the permeate stream. In
this case, countercurrent staging was used to increase impu-
rity removal and/or reduce the required buffer volume (or
flow rate). Possible binding of impurities was neglected, so
all of the terms involving Cb were assumed to be zero; bind-
ing of impurities to the base resin or the bound mAb would
reduce the overall level of impurity removal, which is also
observed in column chromatography. The resin concentration
in the retentate leaving the hollow fiber membrane module
was controlled to be equal to that in the inlet to the static
mixer, i.e., ur 5 uin. For a multistage countercurrent system,
the final impurity removal is given as:17
aðaN 21Þ
R1 ðor R2 Þ5 (10)
aN11 21
a5cSK (11)
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
433
1
K5 (12)
12uave ð12xÞ
uin ðc12Þ
uave 5 (13)
2c12
where R1 ðor R2 Þ is the fractional impurity removal for
wash-1 (or wash-2), N is the number of stages in the given
wash step, c is the ratio of the wash buffer flow rate to the
concentrated slurry flow rate, S is the average sieving coeffi-
cient for the species of interest (S 5 1 for CHO cell proteins
for the 0.45 mm pore size membrane used in the experi-
ments), K is a correction factor that accounts for the pres-
ence of solids in the slurry, uave is the average slurry
volume fraction in the hollow fiber module (evaluated as the
mean of the inlet and outlet values), and x is the liquid frac-
tion contained in the packed (settled) resin bed. For example,
for uin 5 0.25, c 5 3, and uout 5 uin/(c 1 1) 5 0.0625, one
gets uave 50:156: So for S 5 1 and x 5 0.6, Eq. (12) gives
K 5 1.067, with a 5 3.20 (from Eq. (11)), N 5 4, giving
R1 5 99.3% (from Eq. (10)). Note that a system operated
with cocurrent staging would only provide 90.5% impurity
removal using otherwise identical conditions. The total
impurity removal from the combination of the two wash
steps will simply be [12{(12R1)(12R2)}].
The buffer usage (V_ in L/g of mAb) in the wash steps for
a CCTC system, defined as the volume of buffer required
per g of mAb product, is given as:
fc 1c g
V_ 5 w-1 w-2 (14)
cb CF Y
where Y is the overall mAb yield for the CCTC system, CF
is the mAb concentration (titer) in the clarified cell culture
fluid, and cw-1 and cw-2 are the ratios of buffer flow to slurry
flow rates in the wash-1 and wash-2 steps, respectively.
Elution step
The low pH buffer used in the elution step is assumed to
cause complete release of the bound mAb, i.e., Cb,out 5 Cb,r
5 0. The resin concentration in the retentate leaving the hol-
low fiber membrane module was again set equal to that in
the inlet to the static mixer, i.e., ur 5 uin. The use of coun-
tercurrent staging in the elution step provides greater yield

434
Table 1. Key Data Obtained from Binding Experiments
Antibody Resin Size ðlmÞ Qmax (g/L)
mAbl (CCF) 45 26.4
mAb2 (CCF) 45 21.2
mAb3 (Pure) 45 37.8
mAb3 (Pure) 10 27.2
The procedures to estimate rate constants are available in our previous paper (Dutta et al.)19 and are also discussed in Supporting Information.
k1, rate constant for the first stage; k2, rate constant for the second stage; kab, rate constant for "after binder"; ks, rate constant for single stage.
Figure 3. (A) Changes in binding step yield and productivity
with residence time, (B) optimization of yield and
productivity. The mAb titer in the CCF was 7.2 g/L
and the binding capacity of the resin was 26.4 g/L.
The cb was 0.74 corresponding to 81% loading. The
binding step was designed as a single stage system.
The volume fraction of resin slurry was 0.25.
(recovery) of the mAb while minimizing the amount of elu-
tion buffer, which also leads to a greater mAb concentration
in the collected eluate. The yield for a multistage elution
step can be evaluated using Eqs. (10–13) with:
Yelute 5 R
System design
Model calculations were performed for four systems
involving three different monoclonal antibody products: (1)
mAb1 in CCF using Poros 45 mm resin, (2) mAb2 in CCF
using Poros 45 mm resin, (3) pure mAb3 using Poros 45 mm
resin, and (4) pure mAb3 using a prototype 10 mm resin.
The key parameters for the different systems are summarized
in Table 1, with details about the evaluation of these param-
eters given in Supporting Information. The analysis of the
two stage binding step used different binding rate constants
in each stage to account for the fresh resin in the first stage
and the partially saturated resin in the second stage. This is
discussed in more detail in Supporting Information. The pro-
ductivity (P) of the system (in units of g mAb/L resin/h)
was calculated as:
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biotechnol. Prog., 2016, Vol. 32, No. 2
k1 (min21) k2 (min21) kab (min21) ks (min21)
0.64 0.43 0.34 0.49
1.49 0.92 0.96 1.22
- - - 0.40
- - - 1.49
P 5 qelute Celute =RVi ui (16)
where qelute and Celute are the volumetric flow rate and mAb
concentration in the eluate and RViui is the total volume of
resin contained in the CCTC system. Equation (16) can also
be used to calculate the productivity of any given step in the
process by replacing the numerator with the mass flow rate
of product in the stream leaving that step (e.g., the retentate
leaving the after binder in the binding step) with the denomi-
nator evaluated based on the sum of the resin volumes in the
different stages within that particular step.
Results and Discussion
Binding step
The effects of the residence time in the static mixer on the
mAb yield and productivity of the binding step are shown in
Figure 3A for a single stage binding step. The productivity
shown in this plot corresponds to the binding step mAb cap-
ture only, and not the overall productivity of the system.
Calculations were done using the binding parameters for
mAb1 (Table 1) with a resin volume fraction in the inlet
slurry of uin 5 0.25 and an equilibrium loading of 81%
(evaluated by assuming all the mAb is bound to the resin).
The mAb yield increases with increasing residence time as
expected due to the reduction in free (unbound) mAb as the
binding approaches equilibrium. 99% yield is obtained at a
residence time of 7.4 min under these conditions. The pro-
ductivity on the other hand decreases with increasing resi-
dence time since this corresponds to an increase in the
volume of the static mixer (and thus the total resin volume
in the binding step).
The trade-off between the yield and productivity is shown
explicitly in Figure 3B for values of the capture yield >0.95,
which is the target range for high value mAb products. The
productivity decreases with increasing yield due to the high
(15) residence times (and thus large static mixer volumes) needed
to achieve higher mAb binding. However, the binding step
productivity remains more than 200 g/(L h) at mAb yields
up to at least 98%. It is a point to note that the overall sys-
tem productivity will be less than binding step productivity
due to the hold-up volume of resin in the static mixers/hol-
low fiber modules in the other steps. The overall system pro-
ductivity is discussed in the latter part of this article.
The effect of loading on yield and productivity for the
binding step at a fixed residence time of 5 min is shown in
Figure 4. The loading was varied by changing the inlet
slurry flow rate, qs,in. The yield decreases with increasing
loading due to the greater fraction of unbound mAb. Higher
residence times would thus be required at high loadings to
maintain high product recoveries. The productivity increases
with increasing loading due to the reduction in resin volume
contained within the static mixer and hollow fiber membrane
module; this is a direct result of the reduction in qs,in (a

Biotechnol. Prog., 2016, Vol. 32, No. 2
Figure 4. Variation in binding step yield and productivity with
resin loading for a single stage binding step. The
mAb titer in the CCF was 7.2 g/L and binding
capacity of the resin was 26.4 g/L. The residence time
was 5 min, and the volume fraction of resin was 0.25.
smaller volume is needed to achieve the desired residence
time at lower flow rates).
The productivity and yield of the binding step can be
improved by using a two stage binding configuration. A
direct comparison of the performance of the one and two
stage binding steps is shown in Figure 5 for the binding
parameters corresponding to mAb1 (7.2 g/L titer) and mAb2
(0.67 g/L titer) with the loading fixed at 81% for mAb1 and
70% for mAb2 (consistent with the experiments performed
using these mAbs). In both cases, the two stage system
requires less total residence time for a given yield due to the
countercurrent configuration in combination with the added
product recovery in the after binder. The percent increase in
productivity is more pronounced for the low titer mAb2
(Figure 5B) due to greater benefits of countercurrent staging
at lower driving force for binding. The addition of the after
binder in the two stage system is critical to achieving high
product yield in the binding step; a two stage system without
the after binder had very similar performance to a single
stage system at high yield because of the need to minimize
loss in the permeate. For 7.2 g/L titer (mAb1), 81% loading,
u 5 0.25, and 98% binding step capture yield, the total resi-
dence time required for a single stage is 5.75 min. This
decreases slightly to 5.70 min for a two stage system without
the after binder, and it drops to 4.92 min for a two stage sys-
tem with the after binder. Note that removal of the after
binder, keeping the static mixers unchanged, would cause
the mAb yield to drop from 98% to less than 83% due to
loss of unbound mAb in the subsequent washing step.
Further improvements in system productivity can be
achieved by using smaller resin particles (10–20 mm in diam-
eter) which will have much faster binding kinetics due to
reduced time required for pore diffusion. The difference in
binding kinetics for the 10 and 45 mm particle size resins is
shown in Supporting Information. In contrast to column
chromatography, in which the pressure drop increases dra-
matically when using very small particles, the pressure drop
in the CCTC system is essentially independent of particle
size since the viscosity of the flowing slurry depends almost
entirely on the resin volume fraction. In addition, the resins
used in packed columns are highly crosslinked to provide
the mechanical strength needed to withstand high operating
pressures (>100 psi). Resins with lower crosslinking den-
sities, which could potentially have much higher binding
capacities,20 can be used in CCTC due to low operating
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
435
Figure 5. Comparison of the performances of one stage and
two stage binding steps. (A) Results for mAb1
(titer 5 7.2 g/L) with resin binding capacity of
26.4 g/L, cb 5 0.74 (corresponding to 81% loading),
and resin volume fraction of 0.25. (B) Results for
mAb2 (titer 5 0.67 g/L) with resin binding capacity
of 21.2 g/L, cb 5 3.1 (corresponding to 70% loading),
and resin volume fraction of 0.14. The dotted lines
are to guide the eye.
pressures (<20 psi) and the absence of any compaction
effects. Note that the small particle size resins examined in
this study did not show higher binding capacities, which is
likely due to the challenges in obtaining the required high
pore volume and pore size without further optimization of
the particles.
Figure 6 shows the effects of particle size and binding
capacity on the trade-off between the binding step productiv-
ity and yield for mAb3. Calculations are shown for a single
stage binding step at a fixed loading of 70% assuming an
inlet resin concentration of uin 5 0.25. The productivity
decreases with increasing yield, with the performance curves
for the 10 mm resin lying well above those for the 45 mm
resin. Note that a 10 mm particle with low Qmax 5 27.2 g/L
has better productivity than a 45 mm particle with Qmax 5
37.8 g/L. The use of a small particle size resin with high
binding capacity (Qmax 5 54.4 g/L) can provide a binding
step capture productivity of more than 800 g/(L h) at 98%
mAb yield which is more than five times the productivity for
a standard 45 mm particle.
Wash step
The effects of the buffer flow rate and number of stages
on impurity removal in the wash step are shown in Figure
7. It is important to note that the overall removal of
unbound impurities in the CCTC system will be greater
than that shown in Figure 7 due to additional impurity
clearance in the permeate collected in the binding step.

436
Figure 6. Effect of particle size on binding step productivity
for 10 and 45 mm particles. Qmax for the 45 mm par-
ticle was 37.8 g/L while simulations for the small
particle were done using Qmax 5 27.2, 37.8, and
54.4 g/L. The load was 70% ðcb 5 2.77, 3.85, and
5.54 for Qmax 5 27.2, 37.8 g/L, and 54.4 g/L, respec-
tively). The mAb3 titer was 1.72 g/L with a resin
volume fraction of 0.25.
Also, the impurity removal was calculated assuming negli-
gible binding of impurities to the resin or mAb. Similarly
to column chromatography, any impurities that bind to the
product or the resin (and cannot be removed during the
washing stages) would probably coelute with the mAb in
the elution step, which would reduce the overall level of
impurity removal.
As expected, the impurity removal increases with
increasing wash fluid flow rate (or c), but this also leads to
an increase in buffer usage. Alternatively, one can signifi-
cantly reduce buffer usage by increasing the number of
stages at a fixed impurity removal. For example, a 3-log
(1000-fold) reduction in unbound impurities can be
achieved using two 4 stage wash steps (wash-1 and wash-
2), corresponding to 0.72 L of wash buffer (total of wash-1
and wash-2) per g of product, which is a 150% reduction in
buffer requirements compared to 2 stage wash steps.
Although there are diminishing returns on decreasing buffer
requirements by increasing the number of stages (N), signif-
icant savings can be achieved up to N 5 5 stages for most
cases (Figure 7B), although this would lead to greater flow
path costs because of the additional static mixers and hol-
low fiber modules.
Elution step
The effect of countercurrent staging on product yield for
the elution step for different values of c is shown in Figure
8. Calculations were performed with the resin initially loaded
with mAb at a bound concentration of 21.3 g/L (81% of the
maximum binding capacity). As expected, the elution step
yield increases with an increase in the number of stages, par-
ticularly as N increases from 1 to 4. For low buffer flow
rates, e.g., c 5 1.5, more than seven stages are required to
achieve 98% yield. High yields can also be achieved using
fewer stages by employing a higher buffer flow rate,
although this leads to higher buffer usage and more dilute
product in the elution pool. Note that the mAb concentration
could be readily increased by using an in-line ultrafiltration
of the product stream collected from the elution step. The
buffer usage in the elution step, at a fixed yield of 98%,
decreases from 1.18 L/g with N 5 2 to 0.42 L/g with N 5 4
(Figure 8B).
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biotechnol. Prog., 2016, Vol. 32, No. 2
Figure 7. (A) Effect of c on impurity removal and total wash
buffer usage for a fixed number of stages in wash
steps, (B) effect of number of stages on wash buffer
usage to achieve a fixed impurity removal.
Titer 5 7.2 g/L, cb 5 0.74, x 5 0.6, and u 5 0.25. The
dotted lines are to guide the eye.
System optimization
The overall performance of the CCTC system was deter-
mined based on the combined performance of the binding,
wash, elution, and regeneration steps. Calculations were per-
formed for systems with a single stage binding step having
Ybind 5 Yelute 5 0.98, 3-log impurity removal (in the combi-
nation of the two wash steps), and 98% washout in the strip
and equilibration steps (calculated using Eq. (10)). The resin
concentration (uÞ in the retentate exit was fixed at 25% for
all hollow fiber modules and the product loading was main-
tained at 81%.
The effects of the resin binding capacity (Qmax) on the
buffer requirements (in L of buffer per g of recovered mAb)
are shown in Figure 9. The buffer usage rapidly decreases
with increasing Qmax and increasing number of stages
(assumed to be the same in all steps after binding). For
example, a resin with 25 g/L binding capacity requires
5.8 L/g buffer when using only two stages, but this decreases
to 2.1 L/g for four stages. Although further increases in the
number of stages would lead to a further reduction in buffer
usage, this would also increase the number of hollow fiber
membrane modules and, in turn, the overall cost of the
CCTC system. An increase in Qmax has tremendous benefit
as this will directly decrease buffer usage without the need
to increase the number of stages. For example, at Qmax 5
25 g/L, a four stage system requires 2.1 L of buffer/g of
mAb. This buffer usage can be reduced to 1.2 L/g for a three
stage system with Qmax 5 60 g/L. This shows the
 15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biotechnol. Prog., 2016, Vol. 32, No. 2 437

importance of future smaller particle/higher capacity resins

Figure 8. Effect of number of stages on (A) elution step yield,
(B) elution buffer usage to achieve 98% elution step
yield. Titer 5 7.2 g/L, cb 5 0.74, x 5 0.6, and
u 5 0.25. The dotted lines are to guide the eye.
designed specifically for CCTC technology.
Further reductions in buffer requirements can be achieved
by internal recycling of buffers. For example, the low salt
effluent solution used in wash-2 could be recycled to wash-
1, supplemented by appropriate in-line addition of concen-
trated salt to achieve the target ionic strength in the wash-1
step. Similarly, the buffer solution used in the equilibration
step could be recycled and used as feed to the stripping step
after addition of acid to lower the pH to the desired value.
This type of buffer recycling is relatively straightforward to
implement in the CCTC system since all buffers are used
continuously with steady state flows (in contrast to the peri-
odic/cyclic buffer usage in batch and multicolumn chroma-
tography). The effect of buffer recycling on the overall
buffer usage is shown in Figure 9C. For Qmax 5 25 g/L,
u 5 0.25, the buffer requirement was reduced by 37% (from
2.13 to 1.34 L/g) simply by recycling the wash-2 and equili-
bration buffers. In comparison, a column packed with resin
of Qmax 5 25 g/L, packing density of 0.64, 70% loading,
and 12 column volumes (cv) of buffer (3 cv wash-1, 2 cv
wash-2, 3 cv elution, 2 cv stripping, 2 cv equilibration would
require 1.1 L of buffer/g of mAb at the same yield. It is
important to note that increasing the resin slurry volume
fraction (u) in CCTC will proportionally decrease buffer
usage (and vice versa). However, higher resin concentrations
can reduce the critical flux which will require an increase in
membrane area and may make it more difficult to maintain
steady state operation over longer process times.
The overall productivity of the CCTC system is examined
in Figure 10 for both the high titer (7.2 g/L mAb1) and low

Figure 9. (A) Effect of resin binding capacity and number of stages on total buffer usage assuming 98% yield (or washout) in the
binding, elution, stripping, and equilibration steps with a total of three log combined impurity removal in the wash steps
(using the same number of stages in all the steps except the binding step). (B) Close up of Figure 9A, showing low buffer
usage region. (C) Effect of resin binding capacity and buffer recycle (from wash-2 to wash-1 and from equilibration to strip-
ping) on total buffer usage. Titer 5 7.2 g/L, cb 5 0.74, x 5 0.6, and u 5 0.25. The dotted lines are to guide the eye.

438
Figure 10. (A) Optimization of productivity and yield for a
high titer (7.2 g/L) CCF for cb 5 0.74 with c 5 3.0
in all other steps. The volume fraction of resin was
0.25. Number of stages in wash-1, wash-2, and elu-
tion steps was three, while that for stripping and
equilibration steps was two. The binding step con-
tained only a single stage. (B) Optimization of pro-
ductivity and yield for a low titer (0.67 g/L) CCF
for cb 5 3.1 with c 5 3.0 in all other steps. The vol-
ume fraction of resin was 0.14. Number of stages in
wash-1, wash-2, and elution steps was three, while
that for stripping and equilibration steps was two.
The binding step was designed with a two stage
system. The dotted lines are to guide the eye.
titer (0.67 g/L mAb2) products. In both cases, the wash-1,
wash-2, and elution steps employed three stages while the strip-
ping and equilibration steps had two stages, consistent with the
experimental studies performed by Dutta et al.19 Figure 10A
shows results for the high titer mAb, with the resin concentra-
tion in the retentate exit fixed at 25% using a single stage bind-
ing step with a loading of 81%. The overall productivity
decreases with increasing mAb yield due to the additional vol-
ume of the static mixers needed to achieve the required resi-
dence time to capture higher levels of mAb. The calculated
value of the productivity for an overall yield of 96% is 73 g/(L
h) which is 4–5 times greater than a typical packed protein A
column. Scale up modeling shows that a CCTC system with
this productivity would be able to purify a 2,000 L batch of
mAb1 using only 17 L of protein A resin in a single shift. This
will result in tremendous cost savings, both for the expensive
protein A media and the overall process (detailed economic
modeling will be the subject of future publications).
The solid circle symbol in Figure 10A shows the experi-
mental results obtained by Dutta et al. for the above condi-
tions.19 It is important to note that the experimental data used
a 28% resin concentration instead of 25%. The experimental
system also had additional dead volume in the binding step
due to a large amount of extra tubing used to connect the
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biotechnol. Prog., 2016, Vol. 32, No. 2
multiple static mixers that were required to achieve the neces-
sary residence time. The net result is that the experimental
productivity of 67 g/(L h) was approximately 10% lower than
the predicted value of 73 g/(L h). The experimental yield was
also slightly lower (90% compared to 96% in the model) due
to additional losses in the wash-1 and strip steps.
Figure 10B shows results from a similar analysis for the
low titer mAb2. In this case, a two stage binding step with a
loading of 70% and a resin concentration of 14% was used.
The two stage binding was used to increase the productivity
of the system. The productivity for the low titer mAb2 is
significantly lower than that for the high titer mAb1 due to
the greater residence time required to achieve high levels of
binding. The productivity again decreases with increasing
product yield, although the dependence on the yield is con-
siderably weaker than that seen for the high titer mAb1 due
to the use of a two stage binding step for mAb2. In this
case, the overall productivity for 96% product yield is 32 g/
(L h), which is in good agreement with the experimental
value obtained under these conditions [29 g/(L h)].19
Conclusions
The results presented in this study provide the first
detailed analysis of the design and optimization of CCTC
systems for continuous product capture, focusing on the
effects of countercurrent staging and buffer flow rates on
system performance. Countercurrent staging is essential for
achieving high degrees of impurity removal while minimiz-
ing buffer requirements in the CCTC system. Staging also
provides high product yield in the elution step. The calcu-
lated productivity of the CCTC system for a high titer mAb
purification from an industrial cell culture was 73 g of mAb/
L of resin/h with 96% product yield and 3.4-log reduction in
impurity levels; this productivity was in good agreement
with experimental data and is more than four times that of
comparable Protein A columns.
The CCTC system does require higher amounts of buffer
than packed column chromatography; however, buffer usage
in the CCTC system can be significantly reduced through
enhancements in system design and resin characteristics. For
example, a packed column requires 1.1 L of buffer/g of
mAb while a CCTC system with two stages in each step
requires 5.8 L of buffer/g of mAb. However, the use of a
four stage system provides more than a 60% reduction in
buffer with respect to a two stage system (2.1 L/g), while
recycling the buffer in the wash and equilibration steps leads
to an additional 37% reduction (1.3 L/g). In addition, the use
of a smaller particle size resin allows the binding step to
operate with much smaller residence times, which could
potentially provide an additional 150–200% increase in sys-
tem productivity. Further increases in productivity, and
reductions in buffer usage, could be achieved using a protein
A resin with higher binding capacity. Based on the results
presented in Figures 6–9, it should be possible to design a
CCTC system with an overall productivity greater than
190 g/(L h) that uses less than 1.1 L/g of buffer. This is an
order of magnitude greater productivity than current Protein
A columns with similar buffer requirements. Future studies
will examine the impact of key design variables on the eco-
nomics of mAb capture/purification, including comparisons
with more recently developed multicolumn chromatography
systems. The CCTC platform will also be assessed for other

Biotechnol. Prog., 2016, Vol. 32, No. 2
separations, such as anion exchange flow through, ion
exchange capture and mixed-mode separations.
Another aspect that requires thorough investigation is the
scale up of the CCTC system. A large commercial scale sys-
tem would need to employ large static mixers to accommo-
date greater volumetric flow rates. The mixing characteristics
of these large mixers will be studied in detail as part of our
future development efforts. In addition, larger membrane sur-
face area would be scaled based on load and flux require-
ments. On-line process analytical technology (PAT) would
also need to be incorporated as part of any commercial con-
tinuous process. Watson et al.21 recently showed how at line
PAT could be incorporated in the scale up of a microfiltra-
tion unit; a similar approach could potentially be used to
incorporate PAT in a CCTC system.
Acknowledgments
The authors would like to thank Andrew D. Tustian,
Philip A. Ropp, Regeneron Pharmaceuticals, and Fujifilm
Diosynth Biotechnologies for their assistance with the exp-
erimental studies. We would also like to thank the National
Institute of General Medical Sciences (part of NIH, Grant
No.9R44GM108259-02) and Ben Franklin Technology Part-
ners of Central PA for financial support of this project.
LITERATURE CITED
1. Aggarwal S. What’s fueling the biotech engine? Nat Biotechnol.
2007;25:1097–1104.
2. Otto R, Santagostino A, Schrader U. Rapid growth in bio-
pharma: challenges and opportunities. Available at: http://www.
mckinsey.com/insights/health_systems_and_services/rapid_grow-
th_in_biopharma. Accessed November 18, 2015.
3. Protein A capture and its alternatives. Genet Eng News. 2014;
34(18). Available at: http://www.genengnews.com/gen-articles/
protein-a-capture-and-its-alternatives/5334/. Accessed November
18, 2015.
4. Langer ES. Downstream processing: promising technologies are
on the horizon. Biopharm Int. 2010;23(23). Available at: http://
www.biopharminternational.com/downstream-processing-promisi
ng-technologies-are-horizon. Accessed November 18, 2015.
5. Gronemeyer P, Ditz R, Strube J. Trends in upstream and down-
stream process development for antibody manufacturing. Bioen-
gineering. 2014;1:188–212.
6. Kelley B. Very large scale monoclonal antibody purification:
the case for conventional unit operations. Biotechnol Prog.
2007;23:995–1008.
7. Zydney AL. Continuous downstream processing for high value
biological products: a review. Biotechnol Bioeng. 2016;113:
465–475.
15206033, 2016, 2, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/btpr.2250 by CAPES, Wiley Online Library on [15/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
439
8. Zydney AL. Perspectives on integrated continuous bioprocess-
ing—opportunities and challenges. Curr Opin Chem Eng. 2015;
10:8–13.
9. Warikoo V, Godawat R, Brower K, Jain S, Cummings D,
Simons E, Johnson T, Walther J, Yu M, Wright B, McLarty J,
Karey KP, Hwang C, Zhou W, Riske F, Konstantinov K. Inte-
grated continuous production of recombinant therapeutic pro-
teins. Biotechnol Bioeng. 2012;109:3018–3029.
10. Godawat R, Brower K, Jain S, Konstantinov K, Riske F,
Warikoo V. Periodic counter-current chromatography—design
and operational considerations for integrated and continuous
purification of proteins. Biotechnol J. 2012;7:1496–1508.
11. Pollock J, Bolton G, Coffman J, Ho SV, Bracewell DG, Farid
SS. Optimising the design and operation of semi-continuous
affinity chromatography for clinical and commercial manufac-
ture. J Chromatogr A. 2013;1284:17–27.
12. Muller-Sp€ath M, Aumann L, Str€ohlein G, Kornmann H, Valax
P, Delegrange L, Charbaut E, Baer G, Lamproye A, J€ohnck M,
Schulte M, Morbidelli M. Two step capture and purification of
IgG2 using multicolumn countercurrent solvent gradient purifi-
cation (MCSGP). Biotechnol Bioeng. 2010;107:974–983.
13. Langer ES, Rader RA. Introduction to continuous manufactur-
ing: technology landscape and trends. Available at: http://www.
biopharma.com/continuous.pdf. Accessed November 18, 2015.
14. Rosa PAJ, Azevedo AM, Sommerfeld S, Mutter M, B€acker W,
Aires-Barros MR. Continuous purification of antibodies from
cell culture supernatant with aqueous two-phase systems: from
concept to process. Biotechnol J. 2013;8:352–362.
15. Hammerschmidt N, Tscheliessnig A, Sommer R, Helk B,
Jungbauer A. Economics of recombinant antibody production
processes at various scales: industry-standard compared to con-
tinuous precipitation. Biotechnol J. 2014;9:766–775.
16. Hou Y, Brower M, Pollard D, Kanani D, Jacquemart R,
Kachuik B, Stout J. Advective hydrogel membrane chromatogra-
phy for monoclonal antibody purification in bioprocessing. Bio-
technol Prog. 2015;31:974–982.
17. Shinkazh O, Kanani D, Barth M, Long M, Hussain D, Zydney
AL. Countercurrent tangential chromatography for large-scale
protein purification. Biotechnol Bioeng. 2011;108:582–591.
18. Napadensky B, Shinkazh O, Teella A, Zydney AL. Continuous
countercurrent tangential chromatography for monoclonal anti-
body purification. Sep Sci Technol. 2013;48:1289–1297.
19. Dutta AK, Tran T, Napadensky B, Teella A, Brookhart G, Ropp
PA, Zhang AW, Tustian AD, Zydney AL, Shinkazh O. Purifica-
tion of monoclonal antibodies from clarified cell culture fluid
using Protein A capture continuous countercurrent tangential
chromatography. J Biotechnol. 2015;213:54–64.
20. Whitney D, McCoy M, Gordon N, Afeyan N. Characterization
of large-pore polymeric supports for use in perfusion biochro-
matography. J Chromatogr A. 1998;807:165–184.
21. Watson DS, Kerchner KR, Gant SS, Pedersen JW, Hamburger JB,
Ortigosa AD, Potgieter TI. At-line process analytical technology
(PAT) for more efficient scale up of biopharmaceutical microfiltra-
tion unit operations. Biotechnol Prog. 2016;32:108–115.
Manuscript received Dec. 8, 2015, and revision received Feb. 11,
2016.