Journal of Chromatography A, 1393 (2015) 5764

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Purication process of recombinant monoclonal antibodies with

mixed mode chromatography
Sophie Maria a , Gilles Joucla a , Bertrand Garbay a , Wilfrid Dieryck a ,
Anne-Marie Lomenech b , Xavier Santarelli a , Charlotte Cabanne a,
a
Univ Bordeaux, BPRVS, EA 4135, F-33000, Bordeaux, France, Bordeaux INP, BPRVS, EA 4135, F-33000, Bordeaux, France
b
Univ Bordeaux, Centre de Gnomique Fonctionnelle, Plateforme Protome, Bordeaux F-33000, France

a r t i c l e i n f o a b s t r a c t

Article history: An innovative process to purify mAb from CHO cell culture supernatant was developed. This three-
Received 28 January 2015 step process involved two mixed mode resins and an anion exchange membrane. We used a human IgG
Received in revised form 6 March 2015 mixture to determine the optimal conditions for each purication step. Thereafter, the whole process was
Accepted 7 March 2015
evaluated and improved for the purication of a recombinant mAb produced in the supernatant of CHO
Available online 14 March 2015
cells. Once optimized, yield and purity of 88% and 99.9%, respectively were comparable to those obtained
in a conventional process based on a capture step using protein A. In addition, aggregates, HCPs and DNA
Keywords:
levels in the puried fraction were below regulatory specications. Then we used mass spectrometry to
Mixed mode chromatography
Antibody
identify contaminating proteins in the antibody fraction in order to highlight the behavior of HCPs.
Host cell proteins 2015 Elsevier B.V. All rights reserved.
Purication process

1. Introduction
Industrial processes for the purication of recombinant mono-
clonal antibodies (mAbs) produced in CHO cells are usually based
on three successive chromatography steps. The rst consists in
the capture of the recombinant antibody from the culture super-
natant by afnity chromatography using protein A. This technique
has proved to be very efcient in terms of selectivity and purity,
and has become the standard in pharmaceutical companies. How-
ever, a survey indicated that 60% of biopharmaceutical companies
would like to replace the protein A step [1]. Indeed, there are
several technical reasons for this such as protein A leakage, the
formation of protein aggregates during the elution step which is
performed at acidic pH, and the nancial aspect linked to the high
cost of the protein A resins. These problems explain why numerous
studies have attempted to nd alternative technologies to cap-
ture antibodies. Multimodal or mixed-mode chromatography has
recently emerged as a candidate for innovative methods of anti-
body purication. Mixed-mode chromatography was developed in
the late 1950s with hydroxyapatite [2,3]. The following genera-
tions of mixed-mode ligands were developed after 1970 and were
used in numerous applications [47]. In the 80s, these resins were
widely used for the purication of nucleic acids but were curiously
Corresponding author.
E-mail address: charlotte.cabanne@bordeaux-inp.fr (C. Cabanne).
neglected in the protein purication eld. It was not before the
pioneering work of Burton and Harding, who developed a vari-
ant called Hydrophobic Charge Induction Chromatography, that
interest for mixed mode chromatography to purify proteins was
rekindled [8]. They tested numerous ligands having heterocycles
known for their hydrophobicity and demonstrated that the combi-
nation of hydrophobic and ionic interactions led to the emergence
of new selectivity. Since then, mixed-mode chromatography has
been used to evaluate its performances for mAb purication that
does not involve protein A [914].
The second and third chromatography steps used to purify
recombinant antibodies should provide orthogonal modes of inter-
action when compared to the capture step to enable effective
separation of any remaining contaminants (Host Cell Proteins, DNA,
viruses) or unwanted forms of the antibody (aggregates). While the
process uses Protein A for capture, the second and third steps are
respectively cation-exchange chromatography (CEX) and anion-
exchange chromatography (AEX) operated in a owthrough mode.
The CEX step eliminates host cell proteins, aggregates and leached
Protein A, whereas the AEX owthrough step removes contaminat-
ing DNA from the producing cells and achieves further reduction
in host cell protein impurities. This sequence of steps has been
widely adopted as a generic purication scheme for a number of
recombinant mAbs [15,16]. Concerning the AEX step, it is generally
accepted that the owthrough mode is the most effective. Indeed,
at neutral pH, the positively charged antibody does not bind to the
matrix, whereas negatively charged species do [17,18]. However,

http://dx.doi.org/10.1016/j.chroma.2015.03.018
0021-9673/ 2015 Elsevier B.V. All rights reserved.
58 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764
owing to the limitations of the classical chromatography column,
i.e. ow distribution, ow rate limitations and cleaning validation,
membrane chromatography was developed, and anion-exchange
membrane adsorbers are already being used as an alternative to
column chromatography in some mAb manufacturing processes
[19]. In light of this advantage, protocols have been developed for
the use of membrane adsorbers for all purication steps [20].
Our goal was to benet from these recent improvements in
order to develop an innovative process to purify recombinant mAbs
produced by CHO cells. Our strategy was to evaluate a three-step
process based on two successive mixed mode chromatography
steps, followed by an anion-exchange membrane step. Preliminary
experiments to validate the three steps individually were rst per-
formed using a human antibody. Thereafter, the whole process was
evaluated for the purication of a recombinant mAb produced by
CHO cells. We paid special attention to the monitoring of con-
taminating proteins and DNA. At each step of the process, DNA
concentration was measured, the global HCPs content was quan-
tied using an ELISA test, and the identity of the contaminating
proteins was determined by mass spectrometry analyses.
2. Material and methods
2.1. Equipment
An Akta Explorer 100 workstation monitored with Unicorn 5.0
(GE Healthcare, Uppsala, Sweden) was used for chromatographic
experiments. Protein detection was performed by measuring UV
absorbance at 280 nm. A HANNA HI8820N conductivity meter and a
HANNA HI2210 pH meter (Tanneries, France) were used to prepare
buffers.
An ACQUITY UPLC system (Waters Corporation, Milford, USA)
was used for analysis.
The buffers and cell culture supernatant were ltered through
0.45 m cellulose membranes (Sartorius Stedim Biotech, Goettin-
gen, Germany) before use.
2.2. Biological material
Two types of samples were used for purication: (i) a human
immunoglobulin G partially puried from Cohn fractions II and
III, which was purchased from Sigma (St Louis, MO, USA) and
(ii) a crude cell culture supernatant from an antibody-producing
CHO cell line (CHO-DP12, ATCC CRL-12445). These cells pro-
duce a humanized anti-IL-8 antibody and have been adapted to
suspension culture in serum-free medium CD CHO (GIBCO Invit-
rogen SARL, Cergy-Pontoise, France) supplemented with 8 mM
L-glutamine [21]. Cell cultures were performed in a 2 l stirred tank
bioreactor with a supply and control tower (DCU, DFC4, Biostat
B plus, Sartorius Stedim Biotech, Aubagne, France) and a super-
vision system (MFCS, Sartorius Stedim Biotech, Aubagne, France).
Dissolved oxygen was set at 50%, pH value at 7.4 and culture was
performed at 37 C. Supernatant was harvested from extended cul-
tures until viability dropped below 92%. Cell counting and viability
were determined with a hemocytometer (Malassez type) by using
the trypan blue exclusion method at 0.2% (w/v) nal dye concen-
tration [22].
2.3. Chemicals
Buffers and solutions were prepared using chemicals of ana-
lytical grade from Sigma (St Louis, MO, USA). Tris-HCl, sodium
phosphate, sodium citrate, sodium acetate and citric acid were
prepared at desired pH and conductivity.
2.4. Liquid chromatographic resins and columns
We used three types of chromatographic resin: (i) HEA HyperCel
(Pall Life Science, Saint Germain en Laye, France); (ii) Capto MMC
ImpRes (GE Healthcare, Uppsala, Sweden) packed in columns Tri-
corn GE Healthcare (4 ml, 1 cm 5.5 cm) and (iii) HiScreen Capto
MMC column (4.7 ml, 0.77 cm 10 cm) prepacked (GE Healthcare,
Uppsala, Sweden). Height Equivalent to a Theoretical Plate (HETP)
and asymmetry were assessed by injection of 5% acetone (v/v) at
1% of the volume of the column in order to validate the column
package.
We used two types of membrane chromatography: (i) a Sar-
tobind Q 75 and (ii) a Sartobind STIC PA Nano (Sartorius Stedim
Biotech, Aubagne, France). A cleaning protocol was performed after
each purication step according to the manufacturers instructions.
The SEC-UPLC column ACQUITY UPLC BEH200 SEC was pur-
chased from Waters (Mildford, USA).
2.5. Analysis
2.5.1. Antibody aggregates and quantication analysis by
SEC-UPLC
Antibody concentration in the collected fractions was deter-
mined by Size Exclusion Ultra Performance Liquid Chromatography
(SEC-UPLC). Samples of 10 l were injected onto the SEC column
regulated at 30 C. The mobile phase consisted in 0.1 M sodium
sulphate in 0.1 M phosphate buffer pH 6.8. Elution was performed
isocratically at 0.3 ml/min with a run time of 10 min. Antibody peak
areas were determined by integration of the 215 nm signal and
compared to a calibration curve generated with diluted samples of
a known concentration of human polyclonal antibody SLH66 from
Equitech-Bio (Kerrville, Texas, USA).
2.5.2. Host cell protein determination by ELISA
HCPs were quantied with the CHO HCP ELISA kit (#CM015)
from Cygnus Technologies (Southport, NC, USA) according to the
high sensitivity protocol recommended by the manufacturer.
2.5.3. Host cell DNA
The Quant-iTTM dsDNA HS assay kit with the Qubit apparatus
(Invitrogen, Carlsbad, USA) was used to measure DNA concentra-
tion.
2.5.4. Sodium dodecyl sulphate polyacrylamide gel
electrophoresis (SDS-PAGE)
Fractions showing absorbance peaks at 280 nm were analyed
by SDS-PAGE using TGX the Stain-Free FastCast Acrylamide Kit.
Molecular masses were estimated using standard markers (Bio-Rad
Precision Plus Protein Standard). SDS-PAGE gels were revealed by
Stain-Free technology (Bio-Rad).
2.5.5. HCP identication
For each sample, the peptide mixture generated from trypsin
digestion was analyzed by online capillary high-performance liq-
uid chromatography (Dionex, Amsterdam, Netherlands) coupled
to a nanospray LTQ-Orbitrap XL mass spectrometer (Thermo Scien-
tic, Illkirch, France). Data were acquired in a data-dependent mode
alternating, a high-resolution mass spectrometry (MS) scan survey
over the range m/z 3001700, and 10 MS/MS scans in an exclusion
dynamic mode. Identication was carried out by database search
against the Uniprot entries from Cricetulus griseus.
 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764
Fig. 1. Ligands used during the purication process.
3. Results and discussion
3.1. Optimization of each step with a polyclonal antibody
3.1.1. First step: capture
To date, several mixed mode resins have been tested for anti-
body capture [23,24]. Both PPA and HEA HyperCel resins gave good
results in these studies so we examined the following character-
istics to choose the most appropriate one: yield, elimination of
HCPs and pH for elution. The pH value needed to obtain a high
antibody yield was less acidic with HEA HyperCel than with PPA
HyperCel resin. This may be advantageous for the stability of mAb,
given that low pH may trigger mAb aggregation. Nevertheless, the
quantity of HCPs coeluted with the antibody with HEA HyperCel
was slightly higher than that obtained with PPA HyperCel. How-
ever, even though the total amount was higher, the number of HCP
species was lower with HEA HyperCel. We thus selected the latter
for the capture step. It contains a hexyl group that could be involved
in hydrophobic interactions and a protonable amine localized in the
spacer arm (Fig. 1).
Optimal conditions for mAb capture by HEA HyperCel were
previously dened [23]. Briey, 0.1 M sodium phosphate buffer
pH 7.3 and 13 mS/cm were used for equilibration and binding,
then 0.1 M sodium phosphate pH 6.5 and 15 mS/cm were used as
intermediate buffer (wash step) to remove any weakly adsorbed
contaminants. Finally, a 50 mM sodium acetate buffer at pH 4 and
5 mS/cm was used for elution. Using these conditions, we puried a
human IgG prepared extemporaneously in the equilibration buffer.
As expected, one peak was observed during the elution step (data
not shown). SEC-UPLC was then used to determine the yield and
the presence of any aggregates. The pattern obtained after SEC-
UPLC showed that there were more aggregates in the elution peak
(around 80%) than in the starting sample (around 15%). Aggregation
of antibodies at an acid pH is known to occur during purication and
must be limited in order to avoid any undesirable immunological
responses [25]. We consider that there are three ways to overcome
this problem: (i) buffering of the elution fraction with 1 M Tris-HCl
pH 8.5 to raise the pH [26], (ii) addition of arginine to the elution
buffer [27], (iii) the use of sodium citrate as an alternative elution
buffer: [28]. These three possibilities were tested in the column.
First, we buffered the elution fraction by using 1/10 (v/v) of 1 M
Tris-HCl pH 8.5, which resulted in an increase in the pH of the frac-
tion from 4 to 5.4. As expected, the SEC-UPLC pattern indicated
that the percentage of aggregates was reduced to around 10%. The
addition of 1 M arginine-HCl to the elution buffer was also tested,
but the elution peak was smaller and broader. In addition, a second
peak appeared during the cleaning step with 0.1 M citric acid, so
the elution step was not efcient in these conditions and the anti-
body was lost in part. The SEC-UPLC pattern indicated a decrease in
aggregates (around 8%), probably owing to the kosmotropic effect
59
of arginine, but it conrmed the low yield. Finally, we evaluated the
use of 50 mM sodium citrate buffer at pH 4 and 6 mS/cm as elution
buffer. The chromatographic prole showed that the pH transition
that occurred between the wash and the elution steps was not opti-
mal, leading to a broader peak. The wash step was thus removed to
allow a larger pH jump. After this change, we obtained a chromato-
graphic prole similar to that previously obtained after elution by
acetate buffer. The corresponding chromatogram is presented in
Fig. 2. A. The percentage of aggregates was reduced to around 10%,
and the yield was around 97%. In conclusion, the sodium citrate
buffer was chosen because it did not require the use of an additive
like Tris-HCl or arginine. Moreover, citrate is more kosmotropic
than acetate, hence the lower amount of aggregates [29].
3.1.2. Second step: intermediate purication
It has been previously demonstrated that the presence of both
an aromatic group and an amide bond allow protein binding at high
salt concentration [30,31]. In view of these observations, GE Health-
care has commercialized the Capto MMC ligand with multimodal
functionality. It contains a carboxyl group that confers the charac-
teristics of a weak cation exchanger, and a phenyl group that may
be involved in hydrophobic interactions. In addition, other types
of interactions such as hydrogen bonds can occur. The diameter of
the resin beads (75 m) is less than that of the HEA HyperCel resin
(90 m), which leads to better resolution. The structure of Capto
MMC resin is presented in Fig. 1.
Conditions for binding and elution were tested in microplates
as described previously [23,28]. Optimal conditions were found to
be 50 mM sodium citrate buffer pH 4 (6 mS/cm) for equilibration
and 200 mM sodium phosphate buffer at pH 7.5 and 21 mS/cm for
elution. They were tested in columns with the human IgG prepared
extemporaneously in the equilibration buffer (Fig. 2B). Desorption
was obtained by increasing the pH from 4 to 7.5 and by moderately
increasing the conductivity (from 6 to 21 mS/cm), which resulted in
decreasing ionic interactions. However, care must be taken not to
increase the conductivity excessively because this could lead to an
increase in the hydrophobic attraction at the expense of the ionic
repulsion. Another resin has been developed with the same ligand
as the Capto MMC: Capto MMC ImpRes. It has smaller bead size
(40 m) and lower ligand density than Capto MMC in order to
improve resolution and selectivity. Other experiments conrmed
that it reduced aggregates efciently (data not shown). Similar per-
formances were obtained with this resin in our case as the sample
contained few aggregates. Capto MMC is better for productivity
while Capto MMC ImpRes is better for removing aggregates. The
formation of aggregates is mAb- and process-dependent, a factor
that determines which resin should be used.
3.1.3. Third step: polishing
Anion exchange membrane adsorbers are used in the biophar-
maceutical industry in polishing steps in a ow-through mode
[3234,19,35]. Indeed, anion exchange is used to bind numerous
contaminant species: virus particles, host cell proteins and DNA.
Therefore, the goal is to choose conditions where the contaminant
species are negatively charged and thus bind onto the membrane,
whereas the positively charged mAb are able to ow through [36].
The expected performance depends on the type of ligand. The qua-
ternary amine-based ligands are strong anion exchangers but they
are based only on ionic interactions. Consequently, their capacities
are reduced at high ionic strength. On the contrary, the primary
amine-based ligands still have high capacities at high ionic strength
owing to their secondary interactions. The latter type is termed
salt-tolerant.
First, we tested a classical anion exchanger Sartobind Q (Fig. 1).
The optimal buffer for elution of the mAb during the second chro-
matographic step was phosphate buffer at pH 7.5 and 21 ms/cm. At
60 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764
Fig. 2. The three steps chosen for purication of human IgG. (A) First step on the HEA HyperCel resin. (B) Second step on Capto MMC resin. (C) Third step on Sartobind Q
membrane.
this conductivity, salts could compete with contaminants for bind-
ing. Therefore, the human IgG was prepared in this buffer diluted
4-fold in water and then was loaded onto the membrane. The chro-
matogram obtained after anion exchanger Sartobind Q purication
showed a peak in the owthrough fraction as expected (Fig. 2C).
Dilution of the sample before loading onto the membrane could
be avoided by using a salt-tolerant membrane like Sartobind STIC,
which is based on a polyallylamine ligand. In this case, a multivalent
buffer such as a phosphate buffer is not recommended because it
can bind to the ligand and thus decrease the binding capacity. We
chose to condition the antibody directly in a Tris HCl buffer at pH
7.0 before its purication with the STIC membrane. Quantication
of the antibody in the owthrough with SEC-UPLC analysis showed
that the yield was good. The Sartobind STIC membrane could be
used as an efcient polishing step. However, as explained above,
the sample should be previously conditioned in a Tris-HCl buffer,
which implies that the preceding chromatographic step should also
be performed using the same buffer. We tested the second step
with a Tris-HCl buffer for elution, but the chromatographic prole
was not satisfactory with a broad peak for the eluted fraction. An
exchange of buffer could have been foreseen between the second
and the third step, but this would have complicated the process.
We thus decided to use the Sartobind Q for the polishing step, as
dilution was simpler than buffer exchange.
3.2. Purication of a recombinant mAb produced by CHO DP12
3.2.1. Validation of the whole process
The three steps were selected and optimized independently
with human IgG. The process was then validated as a whole with
a mAb from CHO cell culture supernatant. The conditions are sum-
marized in Table 1. Purication steps were chained according to
the conventional purication process. Capture on HEA HyperCel
was performed by injecting 200 ml of supernatant (Fig. 3 A). The
chromatographic prole showed two peaks: a owthrough peak
containing unbound proteins and an elution peak that contained
the antibody. The antibody-containing peak was then applied to
the Capto MMC column (Fig. 3B). The prole showed a small
owthrough fraction and a single elution peak. SEC-UPLC con-
rmed that no antibody was lost in the owthrough fraction. The
antibody-containing elution peak was then diluted to 1/4 with
water before injection onto the Sartobind Q membrane (Fig. 3C).
The prole showed a owthrough peak and a small elution peak.
Fractions were analyzed by SEC-UPLC to measure the amount of
antibody and the percentage of aggregates. HCPs and DNA were also
quantied in each fraction. These data are summarized in Table 2.
Yields of each purication step ranged from 94 to 99%. Purity of
the mAb was greatly improved throughout the process, increasing
from 45% to 99.9%. Concomitantly, HCPs were reduced at each step
with a concentration in the nal fraction product of 15 ppm, which
is an acceptable level. Concerning DNA, 10-fold and 100-fold reduc-
tions were obtained after the rst and third steps, respectively.
SDS-PAGE was performed to analyze the protein content during
the process (Fig. 4). The same volume of sample was loaded in each
well. The two bands corresponding to the heavy and light chains
were predominant in each puried fraction. Some contaminants
Table 1
Steps and conditions of purication.
Binding Elution
HEA HyperCel Na Phosphate (100 mM); Na Citrate (50 mM); pH
pH 7.313 mS/cm 46 mS/cm
Capto MMC Na Citrate (50 mM); pH Na Phosphate (200 mM);
46 mS/cm pH 7.521 mS/cm
Sartobind Q Na Phosphate (50 mM); pH
7.55 mS/cm
 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764 61

Fig. 3. Chromatograms of the three steps for purication of recombinant antibody from CHO cell culture supernatant. (A) First step on HEA HyperCel resin. (B) Second step
on Capto MMC resin. (C) Third step on Sartobind Q membrane.

were detected after the rst step but they fell below detection level
after the second step. Owing to the dilution factor required prior
the Sartobind Q membrane purication step, the intensity of the
mAb bands was lower in this fraction than in the others.

3.2.2. Identication of HCPs

Fig. 4. SDS-PAGE of antibody-containing fractions from the purication of recom-
binant antibody from CHO cell culture supernatant. MM: Precision Plus ProteinTM
Unstained Standards (Biorad). S: culture supernatant. 1: Elution fraction of rst step
on HEA HyperCel resin. 2: Elution fraction of second step on Capto MMC resin. 3:
Flow through fraction of third step on Sartobind Q membrane.
The HCP composition of each fraction, i.e. owthrough, elution
and cleaning fractions, was investigated and identied by mass
spectrometry. The corresponding characteristics (pI, MM) were
retrieved using the Uniprot database and bibliography.
Approximately 400 proteins were identied from the super-
natant. At each step of the process, the number of HCPs signicantly
decreased to 200 and 100 proteins after the rst and second steps,
respectively. At the end of the process, around 30 proteins were
detected.
The distribution of the theoretical isoelectric point (pI) and
grand average of hydropathy (GRAVY) were analyzed (Fig. 5 and
Fig. 6). In the supernatant of the CHO cell culture, the majority of
proteins (56.7%) had isoelectric points between 5 and 8, which cor-
responds to the normal distribution of eukaryotic cellular proteins
[37]. During the purication process, we observed a progressive
enrichment in proteins whose pI ranged from 6 to 8, from 31.3%

Table 2
mAb, HCPs, DNA and aggregate values during the purication process of recombinant antibody from CHO-DP12 cell culture supernatant (for 1 liter of culture).

mAb (mg) Yield (%) Purity (%) HCP (ppm) DNA clearance (log) Aggregates (%)

Supernatant (1 l) 40.2 45 24417 NA 6.5%

Capture HEA HyperCel 37.8 94 98.7 270 2 0.5%
Intermediate Capto MMC 35.5 94 99.3 132 0 0.4%
Polishing Sartobind Q 35.2 99 99.9 15 1 0.3%

DNA clearance is expressed in log 10 reduction; NA: not applicable.

62 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764

Table 3
HCPs identied in mAb fraction at the end of purication (in ppm). Proteins are presented in decreasing order based on their relative amount in the supernatant.

in the supernatant to 71.4%, 83.6%, and 89.2% after the HEA Hyper-
cel, Capto MMC and Sartobind Q steps, respectively In agreement
with this observation, the isoelectric point of the puried mAb was
around 8. With regard to hydrophobicity, the majority of proteins
had GRAVY values in the 0.5 to +0.5 range. The calculated GRAVY
for the puried mAb was 0.311. There was a slight increase in the
proportion of proteins whose GRAVY index was between 0 and 0.5
after the rst two purication steps, ranging from 68% to 82%. These
proteins are slightly hydrophobic and were able to establish some
hydrophobic interactions with the two mixed mode resins. This
could explain why some HCPs were enriched despite carrying the
same charge of the ligand. In addition, it should be remembered that
theoretical pI and GRAVY are calculated with the primary sequence,
so they cannot account for changes that may occur after protein
folding. Different classes of HCPs can be co-puried with the anti-
body: (i) those with an overall charge allowing ionic interactions;
(ii) those with an overall charge unfavorable to ionic interaction but
with sufcient hydrophobicity to counterbalance the ionic repul-
sion; (iii) those with pI and hydrophobicity allowing both ionic and
hydrophobic interactions; and (iv) those which directly interact g
with the antibody and thus are co-puried.
For the sake of clarity, we focused on HCPs that were retrieved
in the nal mAb fraction. Table 3 shows the evolution of their
relative amounts after each purication step. Most of these pro-
teins were already identied in our previous study [23]. Such is
the case of procollagen C-endopeptidase enhancer 1, cathepsin
L1 and peroxiredoxin-1, whereas cathepsin Z was not found as
a main contaminant in the antibody-containing fraction. On the
other hand, we identied other proteins such as cathepsin B, com-
plement C1r-A subcomponent, and legumain. To explain these
differences, the change in the elution buffer between these two
studies, from sodium acetate to sodium citrate, should be taken into
 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764
Fig. 5. Distribution of proteins according to their pI in the supernatant and the
antibody-containing fraction during the purication of recombinant antibody from
CHO cell culture supernatant.
Fig. 6. Distribution of proteins according to their GRAVY value in the supernatant
and the antibody-containing fraction during the purication of recombinant anti-
body from CHO cell culture supernatant.
account. Indeed, buffers can modify protein conformation through
kosmotropic or chaotropic effects [26]. Consequently, this could
modify protein-ligand interaction. This was veried for cathep-
sin Z, which was eluted with the antibody when using sodium
acetate buffer for elution, and in cleaning fractions when sodium
citrate was used. Finally, after the polishing step, cathepsin L1 and
cathepsin B were the most abundant contaminants. They were not
quantitatively abundant in the supernatant, but cathepsin L1 and
cathepsin B were enriched during the rst and second steps, respec-
tively. Thanks to the third step based solely on ionic interactions,
their amount fell to acceptable levels. It has been previously shown
that some HCPs can directly interact with mAbs [38,39]. Such is
the case for glutathione S-transferase P1, complement C1r-A sub-
component, peroxiredoxin-1, lipoprotein lipase and procollagen
C-endopeptidase enhancer 1. In agreement with this nding, we
found these proteins in the puried mAbs fraction. However, our
purication process was efcient enough to lower their amounts
to levels compatible with regulatory recommendations.
4. Conclusion
We describe a three-step process for mAb purication using two
mixed mode resins followed by use of an anion exchange mem-
brane for polishing. The proposed sequence is efcient in terms
of specicity and efciency so it could be used to replace pro-
cesses based on protein A chromatography. The two mixed mode
steps allow good selectivity that is specic to each step. Most of
the main contaminants in the supernatant were removed by these
steps. Finally the third step provided orthogonal selectivity and
good polishing of the remaining DNA and HCPs.
63
The overall yield of the process (88%) and the nal purity of
the mAb (99.9%) are comparable to those of conventional afn-
ity process. Impurities were reduced to levels that meet purity
specications for therapeutic products. This new process based on
chromatography techniques could now be easily implemented in
industrial facilities.
References
[1] E.S. Langer, Bioplan Associates, INC., Pharma Manufacturing, 2009.
[2] A. Tiselius, S. Hjerten, O. Levin, Protein chromatography on calcium phosphate
columns, Arch. Biochem. Biophys. 65 (1956) 132155.
[3] S. Hjerten, Calcium phosphate chromatography of normal human serum and of
electrophoretically isolated serum proteins, Biochim. Biophys. Acta 31 (1959)
216235.
[4] B.H.J. Hofstee, R.B. Dunlap (Eds.), Immobilized Biochemicals and Afnity Chro-
matography, Plenum Publ. Corp., New York, 1974, pp. 4359.
[5] R.J. Yon, R.J. Simmonds, Protein chromatography on adsorbents with hydropho-
bic and ionic groups. Purication of human erythrocyte glycophorin, Biochem.
J. 163 (1977) 397400.
[6] I. Sasaki, H. Gotoh, R. Yamamoto, H. Hasegawa, J. Yamashita, T. Horio,
Hydrophobic-ionic chromatography. Its application to purication of porcine
pancreas enzymes, J. Biochem. 86 (5) (1979) 15371548.
[7] W. Kopaciewicz, M.A. Rounds, F.E. Regnier, Stationary phase contributions to
retention in high-performance anion-exchange protein chromatography: lig-
and density and mixed mode effects, J. Chromatogr. 318 (1985) 157172.
[8] S.C. Burton, D.R.K. Harding, Hydrophobic charge induction chromatography:
salt independent protein adsorption and facile elution with aqueous buffers, J.
Chromatogr. A 814 (1998) 7181.
[9] L. Guerrier, I. Flayeux, E. Boschetti, A dual-mode approach to the selective sep-
aration of antibodies and their fragments, J. Chromatogr. B 755 (2001) 3746.
[10] M.C. Mowry, M. Meagher, L. Smith, J. Marks, A. Subramanian, Production and
purication of a chimeric monoclonal antibody against botulinum neurotoxin
serotype A, Protein Expres. Purif. 37 (2004) 399408.
[11] S. Ghose, B. Hubbard, S.M. Cramer, Evaluation and comparison of alternatives
to Protein A chromatography Mimetic and hydrophobic charge induction chro-
matographic stationary phases, J. Chromatogr. A 1122 (2006) 144152.
[12] H. Bak, O.R.T. Thomas, Evaluation of commercial chromatographic adsorbents
for the direct capture of polyclonal rabbit antibodies from claried antiserum,
J. Chromatogr. B 848 (2007) 116130.
[13] J. Chen, J. Tetrault, Y. Zhang, A. Wasserman, G. Conley, M. DiLeo, E. Haimes, A.E.
Nixon, A. Ley, The distinctive separation attributes of mixed-mode resins and
their application in monoclonal antibody downstream purication process, J.
Chromatogr. A 1217 (2010) 216224.
[14] E. Boschetti, Antibody separation by hydrophobic charge induction chromatog-
raphy, Trends Biotechnol. 20 (2002) 333337.
[15] R.L. Fahrner, H.L. Knudsen, C.D. Basey, W. Galan, D. Feuerhelm, M. Vanderlaan,
G.S. Blank, Industrial purication of pharmaceutical antibodies: development,
operation, and validation of chromatography processes, Biotechnol. Gen. Eng.
Rev. 18 (2001) 301327.
[16] A. Shukla, B. Hubbard, T. Tressel, S. Guhan, D. Low, Downstream processing of
monoclonal antibodiesapplication of platform approaches, J. Chromatogr. B
848 (2007) 2839.
[17] G. Kemp, P. ONeil, Large-Scale Production of Therapeutic Antibodies: Con-
siderations for Optimizing Product Capture and Purication, vol. 1, Plenum
Publishers, New York, 2004.
[18] H. Knudsen, R. Fahrner, Y. Xu, L. Norling, G. Blank, Membrane ion-exchange
chromatography for process-scale antibody purication, J. Chromatogr. A 907
(2001) 145154.
[19] J.X. Zhou, T. Tressel, Basic concepts in Q membrane chromatography for large-
scale antibody production, Biotechnol. Prog. 22 (2006) 341349.
[20] H. Varadaraju, S. Schneiderman, L. Zhang, H. Fong, T.J. Menkhaus, Process and
economic evaluation for monoclonal antibody purication using a membrane-
only process, Biotechnol. Prog. 27 (2011) 12971305.
[21] T.N. Gonzalez, S.R. Leong, L.G. Presta (Genentech), Nucleic acids encoding
humanized anti-IL-8 monoclonal antibodies, US 6025158, 2000.
[22] B.A. Reeb, Dye exclusion test for bone marrow viability, in: E.M. Areman, H.J.
Deeg, R.A. Sacher (Eds.), Bone Marrow and Stem Cell Processing: A Manual of
Current Techniques, F.A. Davis Company, New York, 1992, pp. 403404.
[23] J. Pezzini, G. Joucla, R. Gantier, M. Toueille, A.-M. Lomenech, C. Le Snchal, B.
Garbay, X. Santarelli, C. Cabanne, Antibody capture by mixed-mode chromatog-
raphy: a comprehensive study from determination of optimal purication
conditions to identication of contaminating host cell proteins, J. Chromatogr.
A 1218 (2011) 81978208.
[24] M. Toueille, A. Uzel, J.-F. Depoisier, R. Gantier, Designing new monoclonal anti-
body purication processes using mixed-mode chromatography sorbents, J.
Chromatogr. B 879 (2011) 836843.
[25] A.S. Rosenberg, Effects of protein aggregates: an immunologic perspective,
AAPS J. 8 (2006) 501507.
[26] S. Ugwu, S. Apte, The effect of buffers on protein conformational stability,
Pharm. Technol. (2004) 86113.
[27] H. Hamada, T. Arakawa, K. Shiraki, Effect of additives on protein aggregation,
Curr. Pharm. Biotechnol. 10 (2009) 400407.
64 S. Maria et al. / J. Chromatogr. A 1393 (2015) 5764
[28] G. Joucla, C. Le Snchal, M. Bgorre, B. Garbay, X. Santarelli, C. Cabanne, Cation
exchange versus multimodal cation exchange resins for antibody capture from
CHO supernatants: identication of contaminating Host Cell Proteins by mass
spectrometry, J. Chromatogr. B 942943 (2013) 126133.
[29] A.A. Raibekas, E.J. Bures, C.C. Siska, Ta. Kohno, R.F. Latypov, B.A. Kerwin, Anion
binding and controlled aggregation of human interleukin-1 receptor antago-
nist, Biochemistry 44 (2005) 98719879.
[30] B.-L. Johansson, M. Belew, S. Eriksson, G. Glad, O. Lind, J.-L. Maloisel, N. Norrman,
Preparation and characterization of prototypes for multi-modal separation
media aimed for capture of negatively charged biomolecules at high salt con-
ditions, J. Chromatogr. A 1016 (2003) 3549.
[31] T. Yang, G. Malmquist, B.-L. Johansson, J.-L. Maloisel, S. Cramer, Evaluation of
multi-modal high salt binding ion exchange material, J. Chromatogr. A 1157
(2007) 171177.
[32] C. Boi, Membrane adsorbers as purication tools for monoclonal antibody
purication, J. Chromatogr. B 848 (2007) 1927.
[33] H.L. Knudsen, R.L. Fahrner, Y. Xu, L.A. Norling, G.S. Blank, Membrane
ion-exchange chromatography for process-scale antibody purication, J. Chro-
matogr. A 907 (2001) 145154.
[34] H. Shirataki, C. Sudoh, T. Eshima, Y. Yokoyama, K. Okuyama, Evaluation of
an anion-exchange hollow-ber membrane adsorber containing -ray grafted
glycidyl methacrylate chains, J. Chromatogr. A 1218 (2011) 23812388.
[35] J.X. Zhou, T. Tressel, X. Yang, T. Seewoester, Implementation of advanced
technologies in commercial monoclonal antibody production, Biotechnol. J. 3
(2008) 910.
[36] J. Curling, U. Gottschalk, Process chromatography: ve decades of innovation,
BioPharm. Int. 20 (2007) 1019.
[37] M. Jin, N. Szapiel, J. Zhang, J. Hickey, S. Ghose, Proling of host cell proteins
by two-dimensional difference gel electrophoresis (2D-DIGE): implications
for downstream process development, Biotechnol. Bioeng. 105 (2) (2010)
306316.
[38] N. Aboulaich, W.K. Chung, J.H. Thompson, C. Larkin, D. Robbins, M. Zhu, A novel
approach to monitor clearance of host cell proteins associated with monoclonal
antibodies, Biotechnol. Prog. 30 (2014) 11141124.
[39] N.E. Levy, K.N. Valente, L.H. Choe, K.H. Lee, A.M. Lenhoff, Identication and char-
acterization of host cell protein product-associated impurities in monoclonal
antibody bioprocessing, Biotechnol. Bioeng. 111 (2014) 904912.