Professional Documents
Culture Documents
Introduction
ABSTRACT: We report the use of caprylic acid based impu-
rity precipitation as (1) an alternative method to polishing A number of monoclonal antibodies (mAbs) have become
chromatography techniques commonly used for monoclo- established therapeutic products with more molecules
nal antibody purification and (2) an impurity reduction step of this class being developed and tested in clinical trials
prior to harvesting the bioreactor. This impurity reduction
(Kelley, 2009). Production scale for clinically successful
method was tested with protein A purified antibodies and
with cell culture fluid. First, the operational parameters antibodies can reach several hundred kilograms of drug
influencing precipitation of host cell proteins and high substance per year (Aldington and Bonnerjea, 2007; Kelley,
molecular weight aggregate in protein A pools were investi- 2009). Consequently, the production of mAbs requires
gated. When used as a polishing step, the primary factor considerable investment in process development and
affecting purification and yield was determined to be pH.
manufacturing capability. At such scale, productivity of the
Caprylic acid precipitation was comparable to polishing IEX
chromatography in reducing host cell protein and aggregate manufacturing process becomes increasingly important for
levels. A virus reduction study showed complete clearance of minimizing the cost of goods. Until recently, the productivity
a model retrovirus during caprylic acid precipitation of of the mammalian cell culture titers was recognized as
protein A purified antibody. Caprylic acid mediated impu- limiting the mass throughput of mAb manufacturing
rity precipitation in cell culture showed that the impurity
processes in existing facilities (Birch and Onakunle, 2005).
clearance was generally insensitive to pH and caprylic acid
concentration whereas yield was a function of caprylic acid However, recent advances in cell culture technology and
concentration. Protein A purification of caprylic acid pre- growing demand for mAbs are increasingly focusing on the
cipitated cell culture fluid generated less turbid product pool limitations in purifying large quantities of product in a single
with reduced levels of host cell proteins and high molecular batch (Kelley, 2009; Thommes and Etzel, 2007).
weight aggregate. The results of this study show caprylic acid
precipitation to be an effective purification method that can
Though purification process technology continues to
be incorporated into a production facility with minimal cost adapt to meet new challenges, chromatographic separations
as it utilizes existing tanks and process flow. Eliminating still remain the workhorse of mAb purification (Kelley,
flow through chromatography polishing step can provide 2007; Przybycien et al., 2004). A conventional purification
process intensification by avoiding the process tank volume process includes a protein A capture column followed by
constraints for high titer processes.
two polishing chromatography steps (Shukla et al., 2007).
Biotechnol. Bioeng. 2012;109: 2589–2598.
Occasionally, an anion exchange flow-through column or a
ß 2012 Wiley Periodicals, Inc.
bind-and-elute cation exchange is used as a sole polishing
KEYWORDS: caprylic acid; purification; impurity precipi-
step (Kelley et al., 2008; Moscariello, 2008). Typically, the
tation; mAb process development
polishing steps consist of a combination of ion exchange and
hydrophobic interaction chromatography. The key chal-
lenges in existing manufacturing facilities, imposed by
Cheng Zhang’s present address is Genzyme Corporation, Framingham, MA 01701.
Correspondence to: Y. Brodsky higher pool volumes due to improved productivity of
Received 13 January 2012; Revision received 3 April 2012; Accepted 20 April 2012 high titer cell culture processes, are (1) turning around
Accepted manuscript online 30 April 2012; purification unit operation within the bioreactor turn-
Article first published online 16 May 2012 in Wiley Online Library
(http://onlinelibrary.wiley.com/doi/10.1002/bit.24539/abstract) around time and (2) meeting the in-process pool tank
DOI 10.1002/bit.24539 constraints (Gottschalk, 2008). Increasing column loading
ß 2012 Wiley Periodicals, Inc. Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012 2589
does provide some relief but use of larger columns has its Materials and Methods
limitations. With higher column loading, viscosity at
peak protein concentrations can limit the operating Materials
flow rate (due to pressure limitations) and extend the
processing times. Furthermore, introducing columns The mAbs used in this study (Table I), were expressed in
larger than 2.0 m in diameter, a typical upper limit for CHO cells, and produced at Amgen (Seattle, WA).
industrial scale, is expensive and often not supported by MAbSelect SuRe1 Protein A media was obtained from
current manufacturing facilities (Gottschalk, 2008). GE Healthcare (Uppsala, Sweden). The media was packed in
Alternatively, increased cycles can be performed on Vantage columns (i.d. 1.1 cm) obtained from Millipore
the columns until processing times, buffer and process Corporation (Bedford, MA). 4–25% Tris–glycine gel,
intermediate volumes become rate limiting. Limitations in SDS 2 sample buffer, SDS 10 running buffer, and
scale up of conventional chromatography operations bring standard marker were purchased from Invitrogen (Carlsbad,
about a search for alternative technologies for mAb CA). Silver staining kit was obtained from GE Healthcare.
purification (Low et al., 2007). In addition to high Slide-A-Lyzer dialysis cassettes (3,500 MWCO) were
throughput, a new purification method needs to provide purchased from Pierce (Rockford, IL). TRIS–base, TRIS–
separation power equivalent to a chromatography step, fit HCl, sodium chloride, sodium citrate, sodium sulfate,
within the current design of manufacturing facilities, and be citric acid, sodium acetate, acetic acid, sodium phosphate
at a competitive cost. dibasic, and phosphoric acid were purchased from JT Baker
In this article, caprylic acid precipitation of impurities has (Phillipsburg, NJ). Caprylic acid (Octanoic acid 99%,
been evaluated as a possible alternative to conventional C2875) was purchased from Sigma (St. Louis, MO).
polishing chromatography. Also called octanoic acid,
caprylic acid is an eight-carbon, saturated fatty acid. Its
molecular formula is CH3(CH2)6COOH. This method Equipment
involves selective precipitation of impurities while main- Protein A chromatography was carried out on an AKTA
taining the product in solution. Plasma product manu- Explorer chromatographic system from GE Healthcare with
facturers have effectively precipitated non-immunoglobulin built-in UV, pH and conductivity detectors to monitor
proteins from serum, plasma, and ascites fluid by using column effluent. HPLC analysis was carried out using a
caprylic acid (McKinney and Parkinson, 1987; Parkkinen Waters (Milford, MA) HPLC 2690 Separation module,
et al., 2006; Russo et al., 1983; Steinbuch and Audran, 1969). coupled with Waters 2487 Dual l absorbance detector.
It has also been reported to be an effective agent for Absorbance of protein samples was measured using 8453
inactivating enveloped viruses (Johnston et al., 2003; UV-Visible spectrophotometer with 10 mm path length flow
Korneyeva et al., 2002; Lundblad and Seng, 1991). cell from Agilent (Santa Clara, CA). Centrifugation of the
Caprylic acid precipitation can be implemented in a cell culture supernatant was performed on the Allegra X-15R
bioreactor, prior to cell separation, or post Protein A bench top centrifuge (Beckman Coulter, Fullerton, CA).
capture step. In both cases, the purpose is to eliminate a Turbidity of protein solutions was measured with 2100P
polishing chromatography step and/or to increase overall Turbidimeter (Hach, Loveland, CO).
process robustness.
The objective of this study was to identify operational
conditions for caprylic acid precipitation of impurities
Methods
in mAb process as a means for providing alternatives to
the conventional polishing chromatography step. Two
Protein A Chromatography
possible process scenarios for integration of caprylic
acid precipitation step were investigated. Operational Protein A chromatography was carried out on a 1.1 cm
conditions including pH, caprylic acid concentration, D 10 cm H column. The column was equilibrated with
reaction time, and buffer components were investigated 25 mM tris, 75 mM sodium sulfate, pH 7.4 buffer, loaded to
with respect to yield and purification factors, removal
of host cell proteins (HCP) and high molecular weight Table I. A list of mAbs used in this study along with their IgG subtype and
(HMW) aggregates (includes higher order aggregates isoelectric point (pI).
and dimers), for six protein A purified IgG1 and IgG2
mAb Subtype pI
molecules. Process robustness was evaluated with three
mAbs using a design of experiment (DoE) approach. A A IgG 1 9.1
B IgG 1 8.9
study was also performed to evaluate clearance of
C IgG 2 8.0
Xenotropic Murine Leukemia Virus (xMuLV) with caprylic D IgG 2 6.9
acid precipitation of protein A purified product intermedi- E IgG 2 8.5
ate. Finally, precipitation conditions were tested with cell F IgG 2 8.8
culture fluid, prior to cell separation, for three different G IgG 2 7.0
H IgG 1 9.0
mAbs.
2590 Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012
35 g/L of resin, washed with equilibration buffer, and concentration of 0.5% or 1.0% (v/v). In the negative control,
eluted with 50 mM acetate, pH 3.6. Following elution, the the samples were titrated with 2 M tris base only. Following
column was stripped with 100 mM phosphoric acid and the titration, virus stock was added to the samples to the
regenerated with 0.3 M sodium hydroxide. The eluate was concentration of 5% (v/v). Solutions were gently stirred at
collected, neutralized with 2 M tris base and analyzed with ambient temperature with time samples taken at 10, 30, 60,
HCP ELISA and analytical SEC assays. Turbidity and protein and 120 min. At each time point samples were filtered
concentration were determined following neutralization of with a 0.45 mm filter, neutralized to pH 6–8 with 2 M tris
the elution pool. base and assayed for infectious virus using the TCID50
method. Samples were also frozen to be analyzed by reverse
transcriptase quantitative polymerase chain reaction (RT-
Caprylic Acid Precipitation
qPCR) at a later date.
Precipitation of protein A purified mAb. Samples of protein Reverse transcriptase quantitative polymerase chain reaction
A purified mAbs were dialyzed into 50 mM acetate or (RT-qPCR) assay. Total RNA was extracted from each
50 mM citrate at different pHs between 4.5 and 5.5. Two sample using the QIAamp Viral RNA kit (QIAgen Inc.,
percent caprylic acid emulsion was prepared at the same pHs Chatsworth, CA) and transcribed into cDNA using a high
by dispersing neat caprylic acid in either 50 mM acetate or capacity reverse transcription kit (Applied Biosystems, Foster
50 mM citrate buffers using a vortex followed by a pH City, CA) according to the manufacturers’ protocols. Each
adjustment with 10 N NaOH. Precipitation was initiated by 25 mL qPCR reaction contained 5 mL of cDNA, two primers
adding 10 mL of caprylic acid emulsion to 10 mL of dialyzed which amplify a 76 base pair fragment of the xMuLV env
mAb solution at the same pH. Unless stated otherwise, gene, and a probe as described previously (Shi et al., 2004).
starting caprylic acid concentration in the mixtures was 1% The probe was labeled with a fluorescent reporter dye FAM
(v/v; 60 mM). The mixtures were stirred with a magnetic bar (6-carboxyfluorescein) at the 50 end and a quencher dye
at 170 rpm for 1 h at ambient temperature. Precipitate was TAMRA (6-carboxytetramethylrhodamine) at the 30 end.
removed by filtration through a 0.2 mm filter and discarded. qPCR reactions were performed and analyzed on an ABI
The optimal pH conditions were chosen based on IgG yield Prism 7900 Sequence Detection System (Applied Biosystems,
and product quality. Foster City, CA). A 6 log dilution range of the standard was
Precipitation of cell culture fluid. Cell culture fluid (with assayed along with the samples. The amplification curve of
cells) was titrated to pH 4.9 with 1 M phosphoric acid. the xMuLV standard’s dilution range was used as standard
A 20% caprylic acid emulsion was prepared by dispersing curve to determine the xMuLV particle number in each
neat caprylic acid in Milli-Q water by using a vortex. sample. The assay variability is approximately 0.5 logs.
Precipitation was initiated by addition of 20% caprylic
acid emulsion to titrated cell culture fluid. Unless stated
Analysis
otherwise, starting caprylic acid concentration and pH in the
precipitation mixture was 1% caprylic acid at pH 4.7. The Host cell protein analysis. Host cell protein (Chinese
mixture was stirred with a magnetic bar at 170 rpm for hamster ovary cell protein, or HCP) levels were analyzed
1 h at ambient temperature. Precipitate was removed by by an enzyme-linked immunosorbent (ELISA) assay
centrifugation at 2,000g for 10 min, followed by filtration developed at Amgen.
through a 0.2 mm filter. The supernatant was then titrated Size exclusion chromatography. HMW (high molecular
to pH 7.0 with 2 M tris base and protein A purified. weight) was quantified by analytical SEC with a G3000
SWXL column (4.6 mm D 300 mm L) from Tosoh
Biosciences (Montgomeryville, PA).
Virus Clearance Study
Antibody concentration determination. The mAb concentra-
The study was conducted at the Biosafety Development tion in harvested cell culture fluid was determined with a
laboratory at Amgen (Seattle, WA). Xenotropic Murine POROS protein A column (2.1 mm D 30 mm L) from
Leukemia Virus stock was prepared by culturing chronically Applied Biosystems, while antibody concentrations in
infected mink lung cells (ATCC CCL-64). The virus was protein A purified samples were determined by absorbance
concentrated by centrifugation (Romanowski et al., 2008). at 280 nm using 8453 UV-Visible spectrophotometer with
The titer of the virus containing solutions was determined 10 mm path length flow cell from Agilent.
using a standard 50% Tissue Culture Infectious Dose Quantitation of residual caprylic acid. The residual caprylic
(TCID50) end-point titration assay with the indicator cell acid in the supernatants or eluate pools was quantified
line PG4 (feline astrocyte, ATCC CRL-2032). The titer of the by RP-HPLC with a Waters Symmetry1 C18 column
virus was calculated using a modified Spearman–Karber (150 mm 4.6 mm, 3.5 mm, WAT 200632). The running
method and is expressed as a Log10 TCID50/mL value. protocol, a linear gradient of 30–80% acetonitrile in 0.1%
Purified mAb samples were prepared at 5, 15, and 25 g/L phosphoric acid over 25 min, is a modified Fliszar method
in 100 mM Acetate, pH 4.3. Combination of 1 M sodium (Fliszar et al., 2006). Injection volume was 20 mL and the
caprylate and 2 M tris base were used to titrate samples to flow rate was 1 mL/min. The absorbance was monitored at
desired experimental pH (4.9, 5.1, or 5.3) and final caprylate 220 nm.
Yield
95%
process parameters. Of the mAbs outlined in Table I, six
were selected to test precipitation conditions, including two
90%
IgG1 mAbs (A and B) and four IgG2 mAbs (C, D, E, and F).
It was previously reported that precipitation of non-
immunoglobulin proteins is best obtained at a slightly 85%
4.4 4.6 4.8 5.0 5.2 5.4 5.6
acidic pH (McKinney and Parkinson, 1987; Russo et al., pH
1983). Preliminary experiments showed that precipitation
2000
pH is a key factor for consistent purification performance. In
this study, pH < 4.5 was not effective in precipitating HCP
while pH > 5.5 was found to have adverse effect on product 1600
Table IIa. Effect of caprylic acid on protein A pool impurity reduction and yield for six mAbs.
mAb Opt pH Feed HCP (ppm) Pool HCP (ppm) HCP removal (%) Feed HMW (%) Pool HMW (%) HMW removal (%) Yield (%)
A 5.4 1,100 71 94 1.1 0.8 27 86
B 5.2 29,000 120 100 2.4 1.3 46 99
C 5.1 1,676 106 94 3.3 1.1 67 90
D 5.3 672 22 97 2.9 2.1 28 96
E 5.0 1,850 85 95 1.7 1.5 12 92
F 5.3 2,632 181 93 2.2 1.0 55 100
2592 Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012
protein-protein association and subsequent protein precip- (Table IIa). Impurity clearance and product yield was
itation. At lower pH non-ionized form becomes dominant comparable for IgG1 and IgG2 molecules. As stated earlier,
(Caprylic acid pKa ¼ 4.89) but at the same time leads to a the data discussed previously was investigated in 50 mM
decrease in the solubility of CA. Consequently, concentra- sodium acetate buffer conditions. The impact of process
tion of CA available at lower pH is below the critical variables on caprylic acid precipitation of impurities was the
concentration required for extensive CA binding and same in another commonly used buffer system for protein A
precipitation of mAbs. Although the pH dependent binding purification, 50 mM sodium citrate, as compared to 50 mM
of fatty acids to protein has been linked to protein sodium acetate with selected molecules. No differences in
hydrophobicity (Bull and Breese, 1967), we were unable impurity reduction or yield were observed between the two
to find a correlation between the starting pH and mAb buffer systems (data not shown).
properties such as hydrophobicity (8-anilino-1-naphthalene A modified Fliszar method (Fliszar et al., 2006) was used
sulfonate binding), isoelectric point, and melting tempera- to measure residual caprylic acid in the supernatant protein
ture in this study. A relatively narrow pH window at which solution. Levels of caprylic acid were found to be highly
mAbs begin to precipitate suggests that critical CA dependent on the precipitation pH regardless of mAb
concentration is similar between the antibodies. This is concentration or the starting caprylic acid concentration
not surprising given the highly homologous structure of (investigated for two mAbs). The data presented in Table III
mAbs. Conversely, HCPs generally have lower pI values than show residual caprylic acid is about 0.2% at pH 5.0 and
an antibody and hence could be precipitated with a lower about 0.3% when precipitation is done at pH 5.2. This is
critical CA concentration and at lower pH. Therefore, pH consistent with the solubility profile of caprylic acid in an
constraint on the lower side of the range was driven by aqueous solution. Moreover, CEX purification of precipi-
variability in precipitation of contaminants. Results from tated mAb solution reduces residual caprylic acid to
the screening experiment showed considerable differences undetectable levels (LOD ¼ 0.025%). Historically, sodium
in pH at which HCP starts to precipitate for six mAbs tested caprylate (salt form of CA) has been used in preparation of
(pH 4.5–5.1). Previous work (Shukla and Hinckley, 2008) therapeutic human serum albumin for over 50 years
had suggested that the HCPs were mostly associated with (Korneyeva et al., 2002). From a safety standpoint, the
mAbs, which was the likely reason for the difference in pH at intravenous administration of large volumes of albumin
which the complex starts to precipitate. Differences in host formulated in 4 mM (0.06%) caprylic acid over an
cell protein species due to varied production cell lines for the extended period of time shows low risk for implementation
mAbs were also a likely contributing factor. of caprylic acid containing step in the manufacturing
Table IIa shows the reduction in HCP achieved with six process.
mAbs at the optimized pH. For all cases, levels of HCP were
reduced below 200 ppm, regardless of the starting concen-
tration. These results are at least as good as the clearance
Robustness Study
achieved by mAb specific IEX chromatography platform
purification process (Table IIb). Besides considerable Design of experiment (DoE) approach was used to set up a
clearance of HCP, caprylic acid precipitation was effective 2 2 full factorial experiment with two center points to
in reducing aggregate levels for all six antibodies (Table IIa). investigate the interaction effect between pH and concen-
Extent of HMW aggregate precipitation varied from tration of caprylic acid. Three mAbs were selected for this
molecule to molecule; reduction was observed in dimer study, mAb B (IgG1) and mAbs C and D (IgG2). Conditions
as well as higher order multimers. Overall HMW aggregates chosen for this experiment were intended to mimic a
levels achieved with caprylic acid precipitation step were representative range for a scale-up process. Caprylic acid
comparable to the clearance obtained with the polishing concentrations studied were 0.5% and 1.5% with a center
chromatography step (Table IIb). Higher order aggregates point at 1.0%. Based on previous screening experiments, for
were more susceptible to CA precipitation compared to mAbs B and D, pH points were set at pH 4.9 and pH 5.3 with
dimers, likely due to its lower solubility. In all the cases, a center point at pH 5.1. For mAb C, pH 5.0 and pH 5.3 with
product yield at the optimal pH condition remained a center point at pH 5.15 were chosen based on a narrow
above 90% with the exception of mAb A that was at 86% operating window suggested by the screening experiment.
Table IIb. Impurity reduction and yield for six mAbs obtained on mAb specific IEX chromatography platform purification process.
mAb Polishing step Feed HCP (ppm) Pool HCP (ppm) HCP removal (%) Feed HMW (%) Pool HMW (%) HMW removal (%) Yield (%)
A AEX 811 87 89 1.3 1.2 8 99
B CEX 34,000 5,000 85 2.2 1.3 41 93
C CEX 1,600 150 91 4.5 3.5 22 90
D CEX 529 39 93 1.8 1.5 17 96
E CEX 2,559 92 96 1.3 1.1 15 94
F CEX 2,800 325 88 3.2 1.4 56 90
2594 Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012
Table V. Clearance of xMuLV in purified solutions of mAb B and F.
mAb Viral clearance condition mAb (g/L) Infectivity (Log10) RT-qPCR (Log10)
B pH 4.9, 1% CA 15 3.4 a
4.9
pH 5.1, 0.5% CA 15 4.7a 5.5
pH 5.1, 1% CA 15 4.2a 5.3
pH 5.3, 1% CA 15 3.4a 6.4
F pH 5.1, 1% CA 5 2.8 –
pH 5.1, 1% CA 15 3.0 –
pH 5.1, 1% CA 25 3.0 –
a
Assayed a large volume sample.
can be felt in the downstream unit operations as well of cell culture fluid (titration to pH 4.7 with phosphoric
(Yigzaw et al., 2006). Caprylic acid can be used to induce acid) versus lowering the pH and caprylic acid addition.
flocculation of impurities and cellular debris in the cell Data in Table VII show yields as well as HCP and HMW
culture fluid, while maintaining the mAb in the supernatant, aggregate levels in the protein A pool resulting from
in order to enhance removal of precipitated solids by disc- precipitation of impurities in the cell culture fluid. As can be
stack centrifugation. This method is more efficient and less gathered from the table, a higher fold reduction in HCP and
expensive than super sizing depth filters (Yigzaw et al., HMW aggregates is achieved with caprylic acid precipitation
2008). Precipitation of cell culture fluid was evaluated with when compared to acid precipitation or un-treated control.
mAbs C, F, and G to investigate how it affects the An important factor for consideration of caprylic acid
purification process. Precipitated cell culture fluid was precipitation step is a modest loss of product due to co-
protein A purified and the resultant elution pools were precipitation in cell culture fluid (yield 95%). However, it
analyzed for select product quality. Preliminary experiments is desirable to decrease the impurity burden over the
with mAb C, similar to the screening experiments with protein A step with a modest yield loss (5%). This will
protein A purified mAb solutions, were performed using result in superior filter throughput and a longer protein A
select concentration of caprylic acid and precipitation pH resin lifetime, both of which would contribute to a decrease
for cell culture fluid. Precipitation with 1% caprylic acid was in cost of goods. A direct observation of impurity reduction
sufficient to achieve HCP and HMW aggregate clearance due to flocculation of cell culture fluid can be made in the
with an acceptable yield (Table VI). However, unlike the case of mAb C and mAb F. When the clarified cell culture
protein A purified solutions, the output product quality fluid is flocculated with caprylic acid and the resultant
parameters of cell culture fluid precipitation had minimal supernatant samples analyzed by SDS–PAGE (Fig. 2), a
dependence on pH at caprylic acid concentration of 1%. noticeable reduction in high molecular weight species, along
Consequently, pH 4.7 was selected to be used in subsequent with a multitude of other impurities, is observed. This is also
experiments because of slightly higher yields. Similar pH is substantiated by direct measurement of HCP levels in
reported in the literature to be favorable for caprylic acid harvest cell culture fluid (see Fig. 2 caption). For both mAbs,
precipitation of serum and ascites fluid (McKinney and C and F, HCP levels in the protein A elution pools are
Parkinson, 1987). To further examine the effect of caprylic significantly lower when caprylic acid is used (Table VII).
acid flocculation of cell culture fluid on impurity reduction For mAb F, it appears that the precipitation of protein A
and process yield, three mAbs (C, F, and G) were used. This purified sample (Table IIa) is more effective in reducing
experiment compared the impact of simply lowering the pH HCP level than precipitation of cell culture fluid—181 ppm
versus 569 ppm.
Table VI. Effect of caprylic acid (CA) concentration (at a fixed pH) and
flocculation pH (at a fixed CA concentration) on yield and impurity
reduction for mAb C. Table VII. HCP and HMW aggregate levels in the protein A elution pool
after purification of un-treated feed (control), acid precipitated (AP), and
CA (%) (at pH 4.7) Yield (%) HCP (ppm) HMW (%) caprylic acid (CA) treated cell culture fluid for mAbs C, F, and G. AP and
CA precipitation performed at pH 4.7.
Feed – 6,948 1.6
0.5 98 470 1.0 mAb Condition Floc Yield (%) HCP (ppm) HMW (%)
1.0 93 490 0.6
1.5 79 600 0.7 C Feed – 756 3.1
AP 98 2,197 4.9
1.0% CA 90 137 0.7
Floc pH (at 1% CA) Yield (%) HCP (ppm) HMW (%) F Feed – 1,372 3.8
Feed – 6,948 1.6 AP 94 1,293 3.3
4.3 93 510 0.5 1.0% CA 92 569 1.9
4.7 94 470 0.6 G Feed – 2,729 2.0
5.1 97 780 1.0 AP 100 1,073 2.0
5.5 96 570 1.1 1.0% CA 101 111 1.2
1000
100
Turbidity (NTU)
10
1
4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5
Neutralization pH
Figure 3. Turbidity of mAb C protein A pool neutralized to different pH after caprylic acid treatment of (A) protein A pool or (B) cell culture fluid. Legend: Untreated cell
culture/protein A pool (*), acid precipitated cell culture fluid (^), caprylic acid treated cell culture fluid (&), and caprylic acid treated protein A pool (~).
2596 Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012
capture column requires neutralization of the elution pool area and (2) enabling better facility fit for high cell density/
to a desirable precipitation pH. The final caprylate titer process into existing manufacturing footprints. The
concentration was obtained by using the appropriate ratio current study did not evaluate centrifuge performance but
of tris base to sodium caprylate. The dilution that occurred instead concentrated on demonstrating the effect caprylic
during the titration was minimal due to high concentration acid precipitation has on product purity. Implementation of
of sodium caprylate and tris base (concentration > 2 M). caprylic acid precipitation step in the bioreactor is suitable
Moderate mixing of protein A elution pool during addition from the operational stand point since it does not require an
of sodium caprylate solution was sufficient to form the additional step for precipitate removal. In addition, analysis
requisite emulsion. This method also provided a tight of protein A elution pools revealed no measurable caprylic
control over pH of precipitation reaction. Sodium caprylate acid when precipitated CCF was used (LOD ¼ 0.025%).
was also used for precipitation of cell culture fluid. In this Evaluation of centrifuge and depth filter performance for
case, cell culture fluid was first titrated to pH 4.2 with 1.0 M harvest clarification is an important next step toward
citric acid and then 2.0 M sodium caprylate solution was incorporation of the precipitation step in the purification
used to titrate up to pH 4.7. Two methods of precipitant process.
addition were evaluated using cell culture fluid of mAbs F
and G with 1% caprylic acid or sodium caprylate. For both
mAbs precipitation yield and protein A pool product quality
were comparable between the two methods. For mAb F,
Conclusion
HCP concentration in the protein A pool were reduced from Improvements in the production titers of the mAb cell
1,372 ppm in the non-precipitated control sample to 569 culture process raise concerns about purification process
and 680 ppm for caprylic acid and sodium caprylate bottlenecks in the existing facility resulting in low plant
precipitation method, respectively. For the corresponding throughput and/or poor facility fit using conventional
samples, the HMW aggregate was reduced from 3.8% to purification methods. At commercial scale with multiple
1.9% and 2.3%, respectively. For mAb G, HCP concentra- bioreactor configurations needing swift turnaround, large
tion in the protein A pool were reduced from 2,729 to 111 process volumes associated with flow-through polishing
and 128 ppm with caprylic acid and sodium caprylate chromatography causes constraints due to dilution of pool
precipitation method, respectively. For the corresponding volume resulting in process pool tank limitations. At clinical
samples, the HMW aggregate level was reduced from 2% to scale with the desire to operate at smallest column scale for
1.2% for both the precipitation methods. For all conditions, minimizing resin costs, long processing times due to process
product yield was between 92% and 102%. titers and multiple cycles of operations can lead to
The data presented in the previous sections demonstrate throughput limitations. Precipitation of impurities with
potential of caprylic acid precipitation method to become an caprylic acid provides an alternative to polishing chroma-
effective mAb purification step. Implementation of protein tography step for mAb purification. It can be implemented
A pool impurity precipitation with caprylic acid, however, in a bioreactor prior to cell separation or used to precipitate
would require a dedicated depth filtration step to remove the impurities after protein A capture column. Since precipita-
precipitants. In conventional processing, a substantial tion can be carried out in existing tanks in manufacturing
increase in turbidity is often observed upon neutralization operations, there is no need for substantial capital
of protein A elution pool with most mAbs (Yigzaw et al., investments. This method makes it possible to operate at
2006), which requires a depth filtration step as a guard to a very large scale, requiring short processing time and
sterilizing grade absolute filters. As such, the mAb processes minimal buffer preparation. The process volumes required
would not require additional depth filtration steps as the for the addition of sodium caprylate can be managed in a
caprylic acid precipitated pools could be processed using the relatively small (<500 L) disposable bio-bag at commercial
existing bank of depth filters. Turbidity is significantly scale. In the case of precipitating impurities in cell culture
reduced when caprylic acid precipitation is performed as bioreactor, caprylic acid precipitation method provides a
shown in Figure 3, due to reduction in impurities. higher quality feed stream for purification process,
Minimizing turbidity would also allow for greater in- minimizing the burden on the clarifying depth filter train
process pool stability by minimizing effects of hold duration and the following capture step. Precipitation of impurities
on absolute filter throughput. after the protein A column provides clearance of host cell
Further understanding of the effects of caprylic acid proteins and high molecular weight aggregates that is
precipitation of cell culture fluid should be pursued as equivalent or superior to a polishing chromatography step.
having a cleaner feed stream for a mAb purification process Caprylic acid precipitation of the protein A pool leads to
has several benefits. The goal of precipitation step before the complete clearance of xMuLV virus, which can be claimed as
cell separation is to minimize the burden on the harvest an additional orthogonal and robust step for virus
depth filter train and the following capture column by reduction. Productivity is increased by a short process
increasing efficiency of the disc-stack centrifuge operation. development time and a simple scale-up procedure relative
This has a potential to (1) reduce the cost of goods by to a chromatography step. Notably, caprylic acid precipita-
decreasing required depth filter and/or sterilizing grade filter tion is effective for multiple mAb molecules demonstrating
2598 Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012