Applied Microbiology and Biotechnology (2018) 102:1203-1214
`
`
`
`
`
`
`httpsJ/doi.org/10.1007 /s00253-017-8650-5
`
`
`
`
`
`BIOTECHNOLOGICAL PRODUCTS AND PROCESS ENGINEERING
`
`Insights into the generation of monoclonal antibody acidic
`
`
`
`
`
`
`
`
`
`charge variants during Chinese hamster ovary cell cultures
`
`CrossMark
`
`
`
`Hongping Tang 1 • Shiwei Miao 1 • Xintao Zhang 1 • Li Fan 1 • Xuping Liu 1 • Wen-Song Tan 1 • Liang Zhao 1
`
`
`
`
`
`
`
`
`
`Received: 23 September 2017 /Revised: 8 November 2017 / Accepted: 12 November 2017 /Published online: 13 December 2017
`
`
`
`
`
`
`
`
`
`
`
`
`© Springer-Verlag GmbH Germany, part of Springer Nature 2017
`
`Abstract
`Charge variation is one of the most important heterogeneities during monoclonal antibody (mAb) manufacturing and this study
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`presents insights into the generation of acidic charge variants during cell culture processes. Since acidic variants generate both
`
`
`
`
`
`
`
`
`intracellularly and extracellularly, main charge fraction collected by weak cation exchange chromatography (WCX) was incu­
`
`
`
`
`
`bated in harvested cell supernatant (HCS) to simulate and investigate the extracellular process firstly. It is found that the main
`
`
`
`
`
`
`
`
`
`
`fraction was degraded rapidly into acidic variants rather than basic variants extracellularly, and the degradation sites were located
`
`
`
`
`
`
`
`in both Fab and Fe fragments indicated by papain digestion. Besides, certain process parameters were investigated as their
`
`
`
`
`
`
`
`potential roles in the extracellular process. As a result, media composition showed significant influence on degradation while
`
`
`
`
`
`
`
`culture time point did not, suggesting that the extracellular process was a spontaneous process without enzyme catalysis.
`
`
`
`
`
`Additionally, kinetics study reveals that the extracellular process was a pseudo first-order reaction. The Eapp value (21.59 kcal/
`
`
`
`
`
`
`
`
`
`
`
`mo!) estimated from the Arrhenius equation suggests that the extracellular degradation might be mainly attributed to asparagine
`
`
`
`
`
`
`
`
`
`
`deamidation. Furthermore, we established an acidic variants generation model, indicating that the extracellular process plays a
`
`
`
`
`
`
`
`
`
`dominant role in modulating the final acidic variant level. This study provides better understanding for controlling product
`
`heterogeneity in mAb manufacturing.
`
`Keywords
`
`
`
`
`
`
`
`Chinese hamster ovary cells · Monoclonal antibody · Charge heterogeneity · Acidic variants · Asparagine deamidation
`
`
`
`
`
`
`
`
`
`
`
`Introduction
`
`
`
`
`
`2014), as well as formulation storage (Gandhi et al. 2012;
`
`
`
`
`
`
`
`Yin et al. 2013), which finally result in product heterogeneity.
`
`
`
`Heterogeneity in mAbs is represented by charge variation,
`
`
`
`
`Therapeutic monoclonal antibodies (mAbs) derived from
`
`
`
`typically caused by deamidation, isomerization, oxidation,
`
`
`mammalian cells have been used for the treatment of cancer,
`
`
`
`
`N-terrninal pyroglutamic acid, and C-terminal lysine clipping
`
`
`
`
`inflammatory, autoimmune diseases, and other medical con­
`
`
`(Dick et al. 2007; Liu et al. 2008; Luo et al. 2012; Vlasak and
`
`
`
`ditions for nearly three decades (Aggarwal 2014). As glyco­
`
`
`
`proteins, mAbs are susceptible to a variety of post­
`
`Ionescu 2008).
`Generally, acidic charge variants are defined as the variants
`
`
`
`
`
`
`
`
`translational modifications (PTMs) and degradations during
`
`
`
`with more negative charges and elute earlier than the main
`
`
`manufacturing (Aghamohseni et al. 2014; Ivarsson et al.
`
`
`
`
`species during cation exchange chromatography (CEX) anal­
`
`
`
`
`ysis. Numerous modifications have been discovered to form
`
`
`acidic variants, such as deamidation (Vlasak et al. 2009),
`T he online version of this article
`
`
`
`
`Electronic supplementary material
`
`
`
`glycation (Yuk et al. 2011), sialic acid (Santora et al. 1999),
`
`
`
`(https://doi.org/10.1007/s00253-017-8650-5) contains supplementary
`
`
`cysteinylation (Banks et al. 2008), and fragmentation (Zhang
`
`
`
`material, which is available to authorized users.
`
`et al. 2011), as has been reviewed elsewhere (Du et al. 2012).
`
`
`
`Among them, asparagine (Asn) deamidation has been widely
`
`
`
`
`reported as the major cause of acidic variants (Harris et al.
`2001; Vlasak et al. 2009; Yan et al. 2009; Zhang and
`
`
`
`Czupryn 2003). At neutral and basic pH conditions, it pro­
`
`
`ceeds via the formation of a five-member ring intermediate
`State Key Laboratory of Bioreactor Engineering, East China
`
`
`
`
`
`
`
`(succinirnide) and hydrolyzes into a mixture of aspartate and
`
`
`
`
`
`University of Science and Technology, Shanghai 200237, China
`
`
`
`Hongping Tang and Shiwei Miao contributed equally to this work.
`
`
`
`� Wen-Song Tan
`
`wstan@ecust.edu.cn
`
`� Liang Zhao
`zhaoliang@ecust.edu.cn
`
`� Springer
`
`Pfizer v. Genentech
`IPR2017-02019
`
`
`Genentech Exhibit 2048
`
`

`

`1204
`
`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`isoaspartate (Vlasak et al. 2009). Besides, it is worth noting
`that acidic variants have been substantiated to affect the
`
`in vitro and in vivo properties of mAbs. For instance, Asn
`deamidation may lead to the decrease of biological activity,
`and even to a number of immunologic responses (Cleland
`et al. 1992) depending on its location. Numerous reports have
`revealed that deamidation had substantial effects on antigen
`binding and potency when occurred in the Fab region, espe-
`cially in the complementarity-determining region (CDR)
`(Harris et al. 2001; Vlasak et al. 2009; Yan et al. 2009).
`Besides, glycation, a reaction between reducing sugars and
`either the side chain of a lysine or the N—terminal primary
`amine, increases the formation of aggregates (Banks et al.
`2009). Thus, it is critical to control the charge variants at a
`relatively low and consistent level during mAb manufacturing
`and formulation storage.
`Acidic variants generation has already been reported to be
`affected by manufacturing process parameters (Liu et al. 2016).
`In our previous study, it was found that lowering culture tem-
`perature significantly decreased mAb acidic variant level
`(Zhang et al. 2015a), which was supported by other works
`(Abu-Absi et al. 2010; Kishishita et al. 2015). Other studies
`suggested that the acidic variant level was decreased with low-
`ering culture pH (Horvath et al. 2010; Nagashima et al. 2013).
`It was also found that ambient light showed a notable influence
`on mAb acidic variants generation during cell culture
`(Mallaney et al. 2014). Besides, media components, such as
`bioflavonoid, ascorbate, ferric citrate, and ferrous sulfate, were
`also found to be correlated with the level of mAb acidic vari-
`
`ants because of their oxidant or anti-oxidant activity (Hossler
`et al. 2015; Vijayasankaran et al. 2013). In addition, acidic
`variants generation during formulation storage affected by re-
`lated parameters has been also investigated (Gandhi et al. 2012;
`Pace et al. 2013; Yin et al. 2013). Nevertheless, lacking of deep
`insights into the mechanisms and kinetics of acidic variants
`formation during dynamic cell culture process, as well as the
`effects of process parameters, the acidic variants are still hardly
`controlled in mAb manufacturing.
`In this study, we present a case study demonstrating the
`generation of mAb acidic variants during cell culture process
`as described in Fig. 1. Since the acidic variants generation may
`take place inside and outside of the cells, consisting of both
`intracellular and extracellular processes, main charge fraction
`was separated and collected by weak cation exchange chro-
`matography (WCX) and incubated in harvested cell supema—
`tant (HCS) to simulate and investigate the extracellular pro-
`cess separately. Process parameters, which potentially influ-
`ence the extracellular generation process, along with the gen-
`eration kinetics, were investigated to further understand the
`extracellular process and suggest approaches to control the
`variants. Furthermore, an acidic variants generation model
`was established, which provides an insight into the generation
`of acidic variants during cell culture process.
`
`@ Springer
`
`Materials and methods
`
`Materials
`
`A chimeric anti-CD20 mAb (IgGl) (Zhang et al. 2015b)
`expressed by a Chinese hamster ovary (CHO) cell line was
`used in this study. It is a gift kindly provided by Hisun
`Pharmaceutical Co., Ltd. (Hangzhou, China). The mAb was
`produced by a typical fed-batch culture process at 37 °C in
`250-mL shake flasks (Coming, USA) as previously reported
`(Zhang et al. 2015a) and purified by HiTrap Protein A HP
`columns (GE Healthcare, Sweden) according to the manufac-
`turer instructions. The purified mAb was then concentrated
`and buffer-exchanged with 1 X PBS (pH 7.0) to 20 mg/mL
`using 30,000 MWCO (molecular weight cut-oft) Amicon
`Ultracentrifugation Tubes (Millipore, MA, USA). Four basal
`media (BM-1, BM—2, BM—3, BM-4) used were all in-house
`developed in our laboratory. Unless otherwise stated, mate-
`rials used in this study were purchased from Sigma-Aldrich.
`
`Weak cation exchange chromatography
`and collection of charge fractions
`
`Charge variants were separated and collected by weak cation
`exchange chromatography (WCX). An Agilent 1260 HPLC
`system coupled with a ProPac WCX-10, 4.0 X 250 mm, weak
`cation exchange column (Dionex, CA, USA), was used for
`WCX analysis. About 500 ug purified mAb was injected into
`the column over the first 5 min using mobile phase A (10 mM
`Na2HPO4, pH 7.5) at a flow rate of 1 mL/min. Mobile phase B
`(10 mM Na2HPO4 and 100 mM NaCl, pH 7.5) was initially
`applied to the column at 30% and then linearly ramped to 75%
`over 20 min to elute mAb according to charge difference.
`WCX was performed at room temperature and the eluted
`mAb was detected using a UV detector at 280 nm. Acidic
`and basic variants were defined as represented by the peaks
`that were eluted earlier or later than the main peak. Charge
`fractions were collected from the outlet of the UV detector
`
`according to the retention time. After collection, fractions
`from different injections were then pooled, concentrated, and
`buffer-exchanged with 1 X PBS (pH 7.0) to 20 mg/mL using
`30,000 MWCO Amicon Ultracentrifugation Tubes.
`
`MAb incubation
`
`Five milliliters of cell suspension was taken from the shake
`flask daily and centrifuged (10,000g for 5 min) to discard
`cells. The cell supernatant was then injected into an equilibrat-
`ed Protein A affmity column and the flow-through fluid (with-
`out mAb) was collected as harvested cell supernatant (HCS).
`The pH of HCS was adjusted to 7.0 by 1 M HCl. One milliliter
`of main fiaction stock solution (20 mg/mL in PBS as afore-
`mentioned) was mixed with 19 mL HCS or PBS to give a final
`
`

`

`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`Fig. 1 Workflow of the
`experiment
`
`1205
`
`\‘FMain
`
`fife Variant
`
`VAcidic = VExtracellular +VSecretion
`Intracellular process
`
`vlntracellmar
`
` ‘Y’ ‘Y
`W
`WY (\‘Vi
`Y ‘Y‘
`
`
`
`i fir’i E fif‘ E
`tiff} tiff}
`
`;
`.
`3”;
`
`E tacellular
`probess
`
`
`
`fir’fi lfir’
`w‘Y‘
`
`
` .
`‘
`I
`wcx
`Separation
`
`Centrifuge
`
`Sampling
`
`Protein A
`Separation
`
`Added
`
`Supernatant
`
`Supernatant w/o
`mAb (HCS)
`
`E'"""""""“"““""""'§
`I ~
`~ I
`i - _ - i
`i
`Incubation
`E
`i
`i
`l
`I
`E V
`”V i
`
`concentration of 1 mg/mL of mAb. The HCS or PBS with
`1 mg/mL main fraction was then sterile filtered (0.22 urn),
`aliquoted into several vials with air1ight cap and incubated in
`C02 incubators at 29, 32, 35, and 37 °C protected from light,
`respectively. Samples about 200 uL were taken daily, purified
`by a Protein An affmity column, and kept at — 80 °C for
`further analysis.
`
`Papain digestion
`
`Papain digestion was performed according to the procedure
`previously described (Moorhouse et al. 1997), with some mi-
`nor modifications. Briefly, the purified sample was incubated
`at a final concentration of 1 mg/mL in 100 mM Tris-HCl,
`4 mM EDTA, 1 mM cysteine (pH 7.4). The digestion was
`initiated by the addition of papain (diluted to 1 mg/mL with
`water) to give a final protein to enzyme ratio of 10021, the
`digestion volume was 200 uL. The digestion was carried out
`for 2 h at 37 °C and quenched with 5% TFA. Papain-digested
`mAb (consists of Fab and Fc) was then injected into WCX
`column for charge variation detection. The modified WCX
`method for papain-digested mAb was similar to that for intact
`mAb as mentioned above except that a different linear gradi-
`ent elution from 10 to 75% B in 40 min was used.
`
`Antigen-binding affinity
`
`The antigen-binding affinity of mAb was analyzed by cell
`enzyme-linked immune sorbent assay (ELISA). Briefly,
`
`ELISA plates were first blocked with 3% BSA in PBST con-
`taining 0.05% Tween-20. After washing, Raji cells (CD20-
`positive cells) were added to the plates at a concentration of
`3 X 104 cells/well and settled at 4 °C for 16 h. After washing
`and drying, the cells were fixed by adding 0.25% glutaralde—
`hyde in PBS for 10 min. Standard and test samples were both
`diluted to a series of concentrations, added as the primary
`antibody, and incubated for 2 h at 37 °C. HRP—conjugated
`IgG (goat anti-human) was added and incubated for another
`2 h in the dark at 37 °C. Binding signals were visualized using
`TMB substrate, and the light absorbance was measured using
`an ELISA reader at 450 nm. The EC50 was detennined for
`both standard and test samples to calculate the relative binding
`affinity.
`
`Calculations
`
`MAb acidic variants generation during cell culture process is
`assumed to take place both inside and outside of the cells,
`consisting of intracellular and extracellular process. The rates
`of the two processes are Vlntra and VExtm, respectively. It is
`easy to find that the extracellular acidic variants are from both
`the extracellular process and the secretion of acidic variants
`generated inside of the cells (secretion process, ngmfign).
`Based on these assumptions, the rate at which extracellular
`acidic variants concentration is changed is represented by
`the equation:
`
`VACMiC : VExtm + VSecretion
`
`(1)
`
`Q Springer
`
`

`

`1206
`
`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`The extracellular process is assumed to be a non-reversible
`pseudo first-order reaction and the equation can be changed as:
`
`
`d[Pa]
`dt
`
`: k[Pm] —l— Qa
`
`
`dlMl
`dt
`
`(2)
`
`where [Pa] is the extracellular acidic variant concentration, [Pm]
`is the extracellular main species concentration, Q, is the acidic
`variant level when mAb was secreted from cell, [M] is the
`extracellular mAb concentration, k is the rate coefficient for
`
`the extracellular process and is assumed to be a constant,
`k-[Pm] is the rate of the extracellular process, and Qa% is
`the rate of the secretion process. Integration of this equation
`over time yields (from 0 to t):
`
`rwrram:wtmum+mxum
`
`m
`
`where [Pa(t)] is the extracellular acidic variant concentration at
`time t and [Pa(0)] is the extracellular acidic variant concentra-
`tion at time 0. Since [M] is found to be a function of time, the
`equation can be changed as:
`
`[Pa(t)]—[Pa(0)] : k-lg[Pm]-dt + nga-R-dt : [E] + [1]
`
`(4)
`
`where k-II)[Pm]-dt = [E] and [gQa‘R‘dt : [I]. R is the derivative
`of extracellular mAb concentration [M] with respect to time. [E]
`is the accumulative extracellular acidic variants generation (ex-
`tracellular process) and [I] is the accumulative secretion of acidic
`variants (Secretion process). For recombinant CHO cells, almost
`all of the mAbs synthesized are secreted and the intracellular
`mAbs can be ignored. Thus, the [I] value is approximately equal
`to the accumulative intracellular acidic variants generation (intra-
`cellular process).
`
`In Eq. 4, [Pa] and [Pm] were determined by WCX, rate
`coefficient k was estimated by the mAb incubation in HCS,
`mAb concentration [M] was determined by protein A HPLC
`assay, and the derivative R was calculated. The item I; [Pm]dt
`was approximated from the plot of [Pm] against time using the
`multiple-application trapezoidal rule (Yuk et al. 201 1) and [E],
`[I], and Q, values were then determined. The proportion ofthe
`extracellular process in acidic variants generation S is repre-
`sented by the equation:
`
`
`, m
`S‘m+m
`
`Results
`
`(”
`
`Separation and characterization of charge fractions
`
`The WCX profiles of the unfractionated mAb and three charge
`fractions are shown in Fig. 2a. Three distinct areas were noted
`
`@ Springer
`
`as acidic variants (12.5—15 min), main species (15—16 min),
`and basic variants (16—1 8 min), respectively. The
`unfractionated mAb contained 29% acidic variants, 53% main
`
`species, and 18% basic variants. After WCX separation, the
`purity of the three charge fractions were 95.3% (acidic),
`100% (main), and 91.8% Gaasic), respectively. Besides, there
`was no redistribution of the collected main fraction on
`
`reinjecting into the WCX column, suggesting that mAb was
`stable during the routine manipulation.
`The antigen-binding affinities of the charge fiactions were
`analyzed by cell ELISA and the result are shown in Fig. 2b.
`The unfiactionated mAb is assigned a relative binding affinity
`of 100%. The relative binding affinities of the acidic, main, and
`basic fiactions were 55, 132, and 63%, respectively, suggesting
`that the generation of acidic and basic variants led to 59 and 52%
`reduction in the binding affinity, respectively, when compared
`with the undegraded mAb (main fraction). Besides, the high-
`molecular weight substance (HMWS), the low-molecular
`weight substance (LMWS), and the higher order structures of
`the charge fractions were also investigated by size exclusion
`chromatography (SEC), capillary electrophoresis-sodium dode-
`cyl sulfate (CE-SDS), and circular dichroism (CD). Both
`HMWS and LMWS tend to be eluted in the acidic region while
`the secondary/tertiary structures showed no significant differ-
`ence among the three charge fiactions (data not shown), which
`is consistent with the published observation (Khawli et al. 2010).
`
`Extracellular acidic variants generation
`
`To simulate and investigate the extracellular process of acidic
`variants generation, the collected main fraction was incubated in
`HCS (day 12) and PBS with a concentration of 1 mg/mL.
`Figure 3a shows the WCX profiles of the main fiaction after 6-
`day incubation in HCS and PBS at 37 °C. It is found that the main
`fraction was degraded rapidly into acidic variants during the in-
`cubation in both HCS and PBS. About 15.10% main fiaction was
`
`degraded into acidic variants when incubated in PBS. A much
`higher acidic variant level (35.34%) was observed when the main
`fraction was incubated in HCS for 6 days. Additionally, seldom
`basic variants were observed during the incubation, suggesting
`that the basic variants observed during cell culture process in this
`study should be mainly generated intracellularly. Further investi-
`gation showed that the contents of HMWS and LMWS were also
`increased during the incubation (data not shown), which might be
`partially associated with the acidic variants generation.
`To determine the degradation site during the incubation, the
`incubated main fraction was digested by papain and analyzed
`by the modified WCX. Figure 3b shows the WCX profiles of
`the papain-digested main fraction after HCS incubation. Four
`major distinct areas were noted as Fc-acidic variants (13—
`16 min), Fc-main species (16—1 8 min), Fab-acidic variants
`(20—23 min), and Fab-main species (24—26 min), respectively.
`The acidic variant level of each region (Fab and Fe) was
`
`

`

`1207
`
`
`
`16
`
`18
`
`Hi
`
`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`Fig. 2 Charge variants separation
`and characterization. a WCX
`profiles of the mAb charge
`fractions. b Relative antigen—
`bindjng amnities of the charge
`fractions which were normalized
`to the unfractionated mAb. The
`values presented are the average
`from three independent experi—
`ments. *1) < 0.05 relative to the
`main fraction
`
`a
`mfiU
`
`3°
`
`60
`
`Main
`
`Acidic
`
`Basic
`
`4" Basic (91.8%)
`
`20
`
`Main (100%) A
`
`
`
`
`'Unfractionated
`
`12
`14
`
`b
`
`160
`
`A 140
`
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`5 b
`
`IE
`‘3
`DD
`.5
`1:
`.E
`4:
`d.)
`.2
`*5
`
`a m
`
`Acidic
`
`Main
`
`Basic
`
`Charge species
`
`quantified using peak area normalization. It is found that ap-
`proximately 13.55% Fab and 15.35% Fc were degraded into
`acidic variants during HCS incubation, resulting in a total of
`35.34% acidic variants of the intact mAb.
`
`Parameters influencing the extracellular acidic
`variants generation
`
`Acidic variants were believed to be generated rapidly extra-
`cellularly during cell culture process. Process parameters such
`as culture time point, mAb concentration, and media compo-
`sition, which potentially influence the generation process,
`were further investigated.
`
`To investigate the extracellular generation of acidic variants at
`different culture time points, HCSs from day 0 (flesh media), 3,
`6, 9, 12, and 15 were prepared for incubation and the result is
`shown in Fig. 4a. Interestingly, there was no significant diifer-
`ence among the HCSs (including day 0 when incubated in fiesh
`media), suggesting that the HCSs from different culture points
`presented the same effect on the extracellular acidic variants
`generation and the extracellular acidic variants generation should
`be a spontaneous process without enzyme catalysis. However,
`mAb concentration during the cell culture process was increased
`while it was 1 g/L for all the conditions in the cell-flee assay. To
`evaluate the effect of mAb concentration on the extracellular
`
`acidic variants generation, the main fiaction was incubated in
`
`Q Springer
`
`

`

`a
`
`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`1208
`
`Fig. 3 Chromatography analysis
`of the incubated main fraction. a
`WCX profiles ofthe main fraction
`afier incubation in HCS and PBS.
`b Papain—digested WCX profiles
`of the main fraction after HCS
`incubation
`
`
`
`
`
`Acidicvariant(%)
`
`.o9
`
`HCS
`PBS
`Incubation condition
`
`HCS incubated
`
`PBS incubated
`Main fraction
`
`\
`
`\ Fab-Acidic
`
`Fc-A cidic
`\
`
`HCS incubated
`
`Main fraction
`
`Fc-Main Fab-Main |
`
`
`
`
`
`Acidicvariant(%)
`
`5"o
`
`HCS (day 12) with a series of mAb concentrations. Figure 4b
`shows the acidic variants generation during 6-day incubation. As
`can be seen, the acidic variant level was slightly decreased with
`the increase of mAb concentration. Specifically, the acidic vari-
`ant level on day 6 was decreased from 35.4 to 30.9% when the
`mAb concentration was increased from 1 to 5 g/L, but no sig-
`nificant difference was found. Additionally, media composition
`is another factor which potentially impacts acidic variants gen-
`eration. To this end, four basal media (BM) with different com-
`positions developed in our laboratory were used for incubation
`and various acidic variant levels were obtained on day 6 as
`shown in Fig. 40. Specifically, the incubation in BM-l, BM-2,
`BM-3, and BM-4 led to approximately 36.5, 24.5, 25.8, and
`42.3% acidic variant levels, respectively. Media composition is
`a key factor in modulating the extracellular acidic variants gen-
`eration during cell culture process.
`
`Kinetics study of the extracellular acidic variants
`generation
`
`A kinetics study was conducted to further characterize the extra-
`cellular acidic variants generation process. To this end, the main
`
`fiaction was incubated in HCS (day 12) and PBS at 29, 32, 35,
`and 37 °C, sampling was performed daily for WCX analysis.
`Figure 5 shows the time course of charge variant level during the
`incubation. It is found that the main fiaction was degraded into
`acidic variants gradually during the incubation and a higher tem-
`perature led to a higher acidic variant level (Fig. 5 b). The accu-
`mulative acidic variant levels on day 6 were 15.05% (29 0C),
`21.95% (32 °C), 28.54% (35 °C), and 36.02% (37 °C), respec-
`tively. Besides, a linear regression of the logarithm of main spe-
`cies level (%) against incubation time shows that the extracellular
`process was a pseudo first-order reaction with the rate coefficient
`(k) represented by the fitted slope (Fig. 5c). The k values were
`0.0296 day’1 (29 °C), 0.0414 day’1 (32 °C), 0.0566 day’1
`(35 OC), and 0.0764 da ’1 (37 0C), respectively, when incubated
`in HCS. These values were decreased to 0.0125 day’1 (29 OC),
`0.0170 day’1 (32 °C), 0.0219 day’1 (35 °C), and 0.0289 day’1
`(37 0C), respectively, when incubated in PBS (Fig. 5).
`A linear regression was obtained by plotting the logarithm
`of rate coefficient (lnk) against the reciprocal of temperature
`(1/1‘), suggesting the extracellular acidic variants generation
`followed Arrhenius behavior as shown in Fig. 6. Besides, the
`
`fitted slop represents the apparent activation energy (Eapp) of
`
`@ Springer
`
`

`

`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`a
`50.0
`
`40.0
`
`30.0
`
`20.0
`
`10.0
`
`0.0
`
`
`
`
`
`I
`
`3
`
`I
`
`0
`
`I
`
`6
`
`I
`
`I
`
`12
`
`I
`
`15
`
`9
`
`Culture time point (day)
`
`O'
`50.0
`
`
`
`
`
`
`
`(%)
` Acidicvariantcontent
`
`
`
`
`
`
` Acidicvariantcontent(%)
`
`40.0
`
`30.0
`
`20.0
`
`10.0
`
`0.0
`
`|
`
`1
`
`I
`
`I
`
`4
`3
`2
`mAb concentration (g/L)
`
`I
`
`5
`
`I
`
`
`
`
`
`O 50.0
`
`40.0
`
`
`
`30.0
`
`20.0
`
`10.0
`
` Acidicvariantcontent(%)
`
`
`
`
`
`
`
`
`
`0.0
`
`BM-l
`
`BM-2
`
`BM-3
`
`BM-4
`
`Medium
`
`Fig. 4 Acidic variants generation ofthe main fraction when incubated in
`a HCSs harvested from different culture time points, b HCS with a series
`of mAb concentrations, and c four basal media with different
`compositions. The values presented are the average from three
`independent experiments
`
`the reaction. The Eapp value of the generation was approxi-
`mately 10.87 i 0.72 kJ/mol (21.59 i 1.43 kcal/mol) when in-
`cubated in HCS, which was comparable to that when
`
`1209
`
`incubated in PBS (19.51 i 0.94 kcal/mol) within experimental
`error. These results indicate that the mechanism or approach of
`the extracellular process during cell culture process was sim-
`ilar to that in PBS, although the culture supernatant accelerat-
`ed the generation rate (Fig. 3a).
`
`Acidic variants generation during cell culture process
`
`An acidic variants generation model was established to get an
`insight into the acidic variants generation during cell culture
`process, especially the intracellular process. To this end, cell
`growth, mAb production, and charge variant level were inves-
`tigated and the results are shown in Fig. 7. It is found that the
`time course of mAb concentration emerges as an “S-shape”
`curve as shown in Fig. 7b. By polynomial fitting, it was ap-
`proximately regarded as a function of time ([114] = —
`0.0016 :3 +0.0381 :2 +0.01 1+ 0.0404) and its derivative
`was also calculated (R = — 0.0048 t2 + 0.0762 1 + 0.01).
`Figure 7c shows the time course of charge variants during
`the fed-batch culture. Since the mAb concentration at the be-
`
`ginning of the culture (0 h) was too low to analyze charge
`distribution, the first sample (day 0) was taken at 6 h of the
`cell culture process and the extracellular degradation of mAb
`during the first 6 h was negligible (Fig. 5). It is found that the
`acidic variant level was increased from 12.94% (day 0) to
`35.07% (day 16) while the main species was decreased fiom
`61.14% (day 0) to 37.47% (day 16). The high level of acidic
`variant on day 0 (12.94%) suggests the presence of intracel-
`lular generation of acidic variants and was approximately con-
`sidered as the acidic variant content when mAb was secreted
`
`fiom cell on day 0.
`Based on the above conclusions, the assumptions in the
`Materials and Methods section (Eqs. 1—5) are valid and the
`rate coefficient k was regarded as a constant of 0.0764 day’1
`during cell culture process since culture time point and mAb
`concentration showed no significant effects on the extracellu-
`lar process (Fig. 4). The acidic variant level when secreted
`(Q), the accumulative extracellular acidic variants generation
`([E]), the accumulative acidic variants secretion ([1]), and the
`proportion of the extracellular process (S) were determined
`and the results are shown in Fig. 8. It is found that Q, was
`stable at approximately 10% during the first 12 days and was
`then increased rapidly to 31.37% on day 16 (Fig. 8a). Besides,
`the [E] and [1] values shown in Fig. 8b were all increased with
`time while [E] was increased with a much higher rate. At the
`end of the culture, the accumulative extracellular acidic vari-
`
`ants generation and acidic variants secretion were 0.89 and
`0.30 g/L, respectively. Accordingly, the proportion of the ex-
`tracellular process (S) was increased fiom 42% (day 2) to 75%
`(day 16), suggesting that the extracellular process plays a
`dominant role in modulating the final acidic variant level dur-
`ing cell culture process.
`
`Q Springer
`
`

`

`1210
`
`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`
`aINU
`
`140
`
`120
`
`100
`
`80
`
`00
`
`40
`
`20
`
`0
`
`
`
`
`
`
`L
`.
`
`.
`
`.
`
`,
`
`6
`.
`
`,
`
`B
`.
`
`.
`
`.
`
`,
`
`,
`
`10
`.
`
`,
`
`.
`
`
`12
`5
`16
`18
`20 m
`.
`.
`,
`.
`.
`.
`.
`,
`,
`.
`.
`,
`.
`4
`
`b
`
`40.0
`
`\z;
`‘L
`- 30.0
`g
`O
`g
`0
`‘E 20.0
`N
`>
`.2
`o
`E 10.0
`<1
`
`0.0
`
`0
`
`d
`
`20.0
`
`A
`g 16.0
`v
`E
`E 12.0
`«-
`8
`1:
`5 s 0
`s;
`‘
`>
`.2
`<2
`32
`<
`
`4 0
`
`0.0
`
`0
`
`029°C A32°C l35°C 037°C
`
`,
`
`I
`
`o
`
`.
`
`l
`
`A
`
`'
`
`‘
`I
`I
`o
`‘
`.
`‘
`o
`
`3 .
`.
`.
`.
`.
`.
`.
`.
`1
`2
`3
`4
`5
`6
`7
`
`c
`
`4-80
`
`029°C A32°C
`
`l35°C 037°C
`slap” t= .00290 R’= 0. 99
`stopm=.0.0414 Rz= 0.99
`2
`stop3w=.0.0566 R = 0.98
`_ Slnp37D=-0.0764 R1: 0.99
`
`
`
`
`
`4.60
`
`4.40
`
`a
`:
`a)
`‘E
`w
`3
`.2
`g
`9-:
`._
`E
`1:
`V 4.20
`A
`
`.. .
`
`4.00 .
`.
`.
`.
`.
`.
`.
`.
`0
`1
`2
`3
`4
`5
`6
`7
`
`Incubation time (day)
`
`Incubation time (day)
`
`029°C A32°C n35°C <>37°C
`
`0
`
`:1
`A
`
`<>
`
`n
`
`A
`o
`
`°
`n
`0
`A
`g
`0
`a 8
`
`.
`.
`.
`.
`.
`.
`.
`1
`2
`3
`4
`5
`6
`7
`
`o
`
`e
`
`4.70
`
`4-55
`
`4 60
`‘
`4.55
`
`A
`‘5
`3
`g
`g
`o
`.3
`a.
`m
`.5
`g 4.50
`4.45
`z
`'4
`
`029°C A32°C n35°C <>37°C
`
`
`
`smug—0.0125 R’= 0.98
`slop32t~=-0.0170 112:0.99
`Slnp35t=-0.0219 R1: 0.99
`“he
`Slnp37c= -0.0289 R1 = 0.99
`~~~~~o
`
`_______
`
`
`
`A
`4::
`“*0
`
`4.40 .
`.
`.
`.
`.
`.
`.
`.
`0
`1
`2
`3
`4
`5
`6
`7
`
`Incubation time (day)
`
`Incubation time (day)
`
`Fig. 5 Time course of mAb charge distribution during the incubation in
`PBS and HCS at different temperatures. a WCX profiles of the main
`fraction during the incubation in HCS at 29 °C; acidic variant level
`
`during the incubation in b HCS and (1 PBS; main species level on a log
`scale during the incubation in c HCS and e PBS. The values presented are
`the average from three independent experiments
`
`Discussion
`
`Charge variation, especially acidic variation, is one of the
`most important heterogeneities during mAb manufacturing
`since it can substantially affect the safety and efficacy of
`mAbs (Du et al. 2012). The generation of acidic variants in
`formulation solutions has been extensively investigated,
`which helps to well control the variants during formulation
`development and long-term storage (Gandhi et al. 2012; Yin
`
`et al. 2013). However, the generation of acidic variants during
`cell culture process remains rarely known since the complex
`intracellular and extracellular environment. In this study, the
`main charge fraction separated by WCX was incubated in
`HCS to simulate the extracellular process and a generation
`model was established to get insights into the acidic variants
`generation process during cell culture process.
`Cell culture process leads to high levels of acidic variants
`(usually above 20%) and should be well controlled to ensure
`
`@ Springer
`
`

`

`Appl Microbiol Biotechnol (2018) 102:1203—1214
`
`in...“
`
`...._..__$_\N“
`
`-2.0
`
`I
`
`I
`
`i...............~i
`
`_......____......._
`I : Slap,” = -9.522 R2 = 0.992
`(EW=19.51 kcanal )
`
`
`
`3.26
`
`3.28
`
`3.30
`
`3.32
`
`3.20
`
`3.22
`
`3.24
`
`.
`
`= -
`)
`
`n! 14.0
`
`—I— Viable cell density
`
`12.0 ' +Cell viability
`10.0 -
`
`1211
`
`100.0
`
`80 0
`
`EW
`
`A
`
`:
`
`(
`__._.._....._.._
`
`CS:
`= .
`$7
`
`_.__......_..__.__._.._i
`
`8.0 -
`
`6.0 -
`
`4.0 -
`
`
`
`
`
`9‘PaCellviabilty(%)
`
`NPe
`
`0.0
`
`0
`
`.
`2
`
`.
`4
`
`.
`6
`
`.
`8
`
`.
`10
`
`.
`12
`
`.
`14
`
`0.0
`
`16
`
`1000/T (K-l)
`
`Fig. 6 Arrhenius plot for the acidic variants generation during the
`incubation in PBS and HCS. The slop reflects apparent activation
`energy (Eupp) of the degradation reaction The values presented are the
`average from two independent experiments
`
`4.0 -
`
`3.5 -
`
`Culture time (day)
`
`.1
`
`2.0 - Viablecell
`
`
`
`density(106cells/ml)
`
`
`y = -0.0016x3 + 0.038111z + 0.01x + 0.0404
`,
`2=0.999
`”ii i
`
`,1}
`
`,
`
`,i'
`
`(1’
`
`,Il
`
`3.0 -
`
`2.5 —
`
`2.0—
`
`1.5 -
`
`
`
`
`
`mAbconcentration(g/L)
`
`the product quality (Kishishita et al. 2015; Zhang et al. 2015a).
`Fortunately, researches have revealed that this generation pro-
`cess could be effectively modulated by the application of a
`quality by design (QbD) approach (Horvath et al. 2010;
`Nagashirna et al. 2013). Certain critical process parameters
`(CPPs) correlated with the acidic variants generation, such as
`pH (Horvath et al. 2010; Nagashirna et al. 2013), temperature
`(Kishishita et al. 2015; Zhang et al. 2015a), and ambient light
`(Mallaney et al. 2014), have been discovered. In this study, the
`effects of culture time point, mAb concentration, and media
`composition on the acidic variants generation were further in-
`vestigated. It is found that media composition showed remark-
`able effect on the extracellular process (Fig. 4c). Previous stud-
`ies have shown that ferric and bioflavonoid, oxidant and anti-
`
`oxidant, presented opposite effects on the acidic variant gener-
`ation (Hossler et al. 2015). Actually, extensive research efforts
`have demonstrated that medium design has a great potential to
`modulate the quality attributions of mAbs, includi