Short- and long-term effects on mAb-producing CHO cell lines after cryopreservation

Cryopreservation provides the foundation for research, development, and manufacturing operations in the CHO‐based biopharmaceutical industry. Despite its criticality, studies are lacking that explicitly demonstrate that the routine cell banking process and the potential stress and damage during cryopreservation and recovery from thaw have no lasting detrimental effects on CHO cells. Statistics are also scarce on the decline of cell‐specific productivity (Q_p) over time for recombinant CHO cells developed using the glutamine synthetase (GS)‐based methionine sulfoximine (MSX) selection system. To address these gaps, we evaluated the impact of freeze‐thaw on 24 recombinant CHO cell lines (generated by the GS/MSX selection system) using a series of production culture assays. Across the panel of cell lines expressing one of three monoclonal antibodies (mAbs), freeze‐thaw did not result in any significant impact beyond the initial post‐thaw passages. Production cultures sourced from cryopreserved cells and their non‐cryopreserved counterparts yielded similar performance (growth, viability, and productivity), product quality (size, charge, and glycosylation distributions), and flow cytometric profiles (intracellular mAb expression). However, many production cultures yielded lower Q_p at increased cell age: 17 of the 24 cell lines displayed ≥20% Q_p decline after ∼2–3 months of passaging, irrespective of whether the cells were previously cryopreserved. The frequency of Q_p decline underscores the continued need for understanding the underlying mechanisms and for careful clone selection. Because our experiments were designed to decouple the effects of cryopreservation from those of cell age, we could conclusively rule out freeze‐thaw as a cause for Q_p decline. © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 34:463–477, 2018

cryopreservation; freeze‐thaw; CHO cell lines; cell age; antibody

Cryopreservation is an integral part of mammalian cell culture activities. This critical operation generates frozen cell substrates that can be stored and then thawed on demand to inoculate cultures for research, process development, or manufacturing purposes. In biopharmaceutical manufacturing, the creation of well‐characterized cell banks containing frozen cell substrates is a key component of the quality control program.[1] Each cell bank affords a consistent starting point, delivering a safe cell source that has been extensively characterized and tested, for initiating manufacturing campaigns throughout the lifecycle of the biopharmaceutical product.

In recognizing the criticality of cryopreservation to mammalian cell culture activities, Barnes et al.[2] investigated the consequences of cryopreservation on the stability of revived non‐secreting murine myeloma (NS0) cell lines. After monitoring the growth and productivity of four NS0 cell lines expressing the same recombinant protein over ∼4 months, they found that freeze‐thaw did not affect the stability outcome: changes in recombinant protein expression with cell age were independent of cryopreservation. Unfortunately, the impact of cryopreservation on product quality attributes, the cornerstone of cell culture consistency assessments,[3] was not evaluated. In addition, the cryopreservation medium was supplemented with 10% serum, consistent with contemporaneous practices.[2] By contrast, cryopreservation media are now typically serum‐free, as part of an industry‐wide adoption of animal component‐free media to minimize contamination risks from adventitious agents. Removal of serum from the cryopreservation medium may elevate susceptibility of post‐thaw cultures to multiple cell death pathways.[4] Hence the conclusions regarding cryopreservation in the serum‐supplemented background using NS0 cells[2] may not hold for serum‐free cryopreservation of Chinese hamster ovary (CHO) cells, the predominant host for the commercial production of recombinant therapeutic proteins.[5]

Since the landmark study on the effects of cryopreservation using NS0 cells,[2] several studies using CHO cells have shown an association between cryopreservation and loss in productivity. In testing CHO pools for rapid material generation, Nelving et al.[6] observed changes in antibody titers for the CHO pools after cryopreservation. In testing the production stability of recombinant CHO‐K1 cell lines that had undergone glutamine synthetase (GS)‐mediated gene amplification in the presence of methionine sulfoximine (MSX), Jun et al.[7] noted 33–67% loss in cell‐specific productivity (Qp) for four antibody‐expressing cell lines after cryopreservation. However, the potentially causal relationship between cryopreservation and loss in productivity could not be ascertained because cell age was a confounding factor in both studies. A more recent investigation into cell line instability permitted such a causality appraisal; Misaghi et al.[8] found that freeze‐thaw exacerbated loss in productivity for CHO cell lines constitutively expressing a specific antibody. The detrimental effects of apoptotic and necrotic cell death cascades that initiate post‐thaw have been documented for mammalian cultures[4] ; such freeze‐thaw related stresses may be pertinent to recombinant CHO cell lines. Therefore, the return to function for cells after cryopreservation and recovery from thaw may not be as reliable or predictable as the routine reliance on cryopreservation for supporting mammalian cell culture activities would suggest.

Within our CHO cell culture workflow, cell banking typically marks the boundary between two stages of development for a biopharmaceutical product: (1) cell line development, which ends with the cryopreservation of the top cell lines selected by clone screening; and (2) process development, which starts with the thaw of these cell banks for selection of the lead and backup cell lines. Historically, we have encountered loss in productivity for recombinant CHO cell lines between these two stages of development (unpublished observations). Unfortunately, we could not distinguish between the effects of freeze‐thaw and cell age because cryopreservation was confounded with increased cell age in our workflow.

The primary goal for this study was to confirm or refute the causal relationship between cryopreservation and CHO production outcomes. To our knowledge, there is no published research to date that explicitly evaluated the effects of cryopreservation on recombinant CHO cell lines expressing different antibodies. Therefore, the observed associations between cryopreservation and production instability[6] [7] [8] warrant further investigations.

The secondary goal for this study was to quantify the occurrence of Qp decline over time in recombinant CHO cell lines. Production stability data are relatively scarce for recombinant CHO cell lines generated using the GS selectable marker and maintained under MSX selective pressure in the seed train (referred to herein as “GS/MSX‐CHO cell lines”).[5] [7] [9] [10] By contrast, abundant data exist for the production stability of recombinant CHO cell lines generated using the dihydrofolate reductase (DHFR) selectable marker and methotrexate (MTX) selection system, as described in the review by Barnes et al.[11]

Flow cytometric studies using hybridoma cell lines have shown correlations between intracellular antibody content and product titer.[12] [13] [14] In particular, the decline in titer observed in these studies was accompanied by the appearance of a subpopulation of cells with lower intracellular antibody content (as determined by staining with fluorochrome labeled anti‐IgG); this co‐existence of lower and higher antibody‐expressing cells within a culture resulted in two peaks on fluorescence intensity histograms. More recently, Dorai et al.[10] applied a similar flow cytometric approach toward evaluating the stability of monoclonal antibody (mAb) production in recombinant CHO cell lines, and demonstrated its utility in identifying unstable cell lines. They performed flow cytometry on CHO cells that were fixed, permeabilized, and stained with fluorescein isothiocynate (FITC) labeled anti‐immunoglobulin G (IgG) antibody to assess intracellular mAb content. To build upon the foundation laid by Dorai et al.,[10] we evaluated the utility of a similar but simpler flow cytometric approach in our clone screening workflow. In addition, we wanted to determine the effects of cryopreservation on flow cytometric profiles because these profiles represent an aspect of mAb expression that standard production culture assays do not capture. A fluorescence intensity histogram generated by flow cytometry analysis provides a snapshot of the distribution of intracellular mAb content within the population of cells at the time of sampling. By contrast, a standard production culture assay provides the mean population value, such as average Qp, within the culture; such a Qp value represents an overall function of the intracellular expression, cellular secretion, and extracellular accumulation of the mAb product.

To achieve these goals, we performed a comprehensive examination of the cell culture, product quality, and flow cytometric profiles for 24 recombinant GS/MSX‐CHO cell lines using a sequential series of experiments. We designed this series of experiments (outlined by the flow chart in Figure ) to satisfy the following considerations: (1) the experimental design should allow for the decoupling of the effects of cryopreservation from those of cell age; (2) the panel of recombinant CHO cell lines used should be sufficiently broad to cover different mAbs and empower statistical analyses; and (3) the cell age range should cover the typical durations encountered in manufacturing from thaw to harvest. By providing parallel seed train cultures for each cell line—one sourced from cryopreserved cells and the other maintained in continuous long‐term culture without cryopreservation—and by positioning the parallel cultures within a similar time frame for a given experiment (Table ), we could discern the impact of freeze‐thaw from that of cell age.

Cell Age (from time of Cryopreservation) at Completion of Main Cell Culture Activities

1 Time in cryopreserved state was not included in cell age calculations.

• 2 Cryopreserved (C) and non‐cryopreserved (NC) cells were used to source each experiment (production culture assay).
• 3 Population doubling level (PDL) was estimated using the average growth rate of seed train cultures from time of cryopreservation to start of Experiment 3.

Figure summarizes the main steps involved in generating and testing the recombinant CHO cell lines used in Experiments 1, 2, and 3. The CHO‐K1 parental strain used to generate these cell lines had been adapted to suspension growth in chemically defined media. This host was transfected with expression vectors containing the GS gene and the cDNA for the heavy and light chains of the desired recombinant humanized mAb as previously detailed.[15] The transfected cells were passaged every 3–4 days in proprietary chemically defined glutamine‐free medium supplemented with 25 μM MSX to maintain GS selective pressure. Cells with the highest mAb productivity were subjected to single‐cell cloning by limiting dilution and expanded. For each set of cell lines producing a specific mAb (i.e., mAb A, mAb B, or mAb C), the top eight cell lines—ranked based on mAb titer and product quality from shake flask production culture assays—were selected for cryopreservation. These cell lines also met the criterion of exhibiting a single flow cytometry peak via intracellular mAb staining prior to cryopreservation. For the 24 cell lines tested, the first set of eight expressed mAb A, the second set of eight expressed mAb B, and the third set of eight expressed mAb C. The time from transfection to cryopreservation was ∼4.6 months for mAb A cell lines, ∼5.8 months for mAb B cell lines, and ∼3.9 months for mAb C cell lines. All seed train cultures were passaged every 3–4 days at inoculation density of 3–4 × 105 viable cells/mL, and cell density and viability were measured by Vi‐Cell XR (Beckman Coulter). The seed train cultures were maintained in shaken (150 rpm, 25 mm throw), vent‐capped Erlenmeyer flasks (Corning) under a 37°C and 5% CO2 environment.

A continuous long‐term culture of each cell line was maintained in the presence of 25 μM MSX; this constituted the seed train from non‐cryopreserved (NC) cells. For each cell line, cells were taken from the NC seed train for cryopreservation and then an ampoule of cryopreserved cells was thawed in the absence of MSX and subsequently passaged in the presence of 25 μM MSX; this constituted the seed train from cryopreserved (C) cells. Therefore, parallel seed train cultures were maintained for each cell line—one derived from non‐cryopreserved cells and the other from cryopreserved cells—to provide cells for production culture assays in Experiments 1, 2, and 3 (Table ).

Cells used for cryopreservation were obtained from seed train cultures within 3–4 days of subculturing in agitated flasks (150 rpm, 25 mm throw, 37°C with 5% CO2 overlay). After centrifugation (260g, 10 min), cell pellets were resuspended to a final target viable cell concentration (VCC) of ∼3.5 × 107 cells/mL in chilled (2–8°C) chemically defined seed train medium. The cryoprotectant dimethyl sulfoxide was added slowly to the cell suspensions to reach a final concentration of 7.5% (v/v). The cell suspensions were transferred to glass ampoules (Amcor) and sealed using the HS1 ampoule and tube hand sealer (Cozzoli Machine Company). The sealed ampoules were placed in Mr. Frosty containers (Thermo Fisher Scientific) in a Forma 8600 series −86°C freezer (Thermo Fisher Scientific) and transferred after 15–72 h to long‐term storage in liquid nitrogen in an MVE 1500 series −190°C cryogenic freezer (Chart Industries, Inc.).

Cryopreserved cells were recovered by thawing the frozen ampoules rapidly (∼2 minutes) at 37°C, followed by immediate transfer of the thawed contents into pre‐warmed culture medium (37°C) to achieve an inoculation density of 3–5 × 105 cells/mL. The thaw medium did not contain MSX; otherwise it was identical to the chemically defined and glutamine‐free seed train culture medium used for subsequent passages. At the end of the thaw passage (3–4 days), if the VCC was ≥6 × 105 cells/mL, the cells were passaged by diluting with fresh seed train culture medium to 3–5 × 105 viable cells/mL. However, if the VCC was <6 × 105 cells/mL, the culture was centrifuged (830g, 5 min), and the resulting cell pellet was resuspended in 100% fresh seed train culture medium to achieve inoculation density of 3–5 × 105 viable cells/mL for the next passage.

Experiments 1, 2, and 3 represent three sets of production culture assays spaced ∼1 month apart (Table ). Within each experiment, both cryopreserved and non‐cryopreserved cells were used. The preparation, setup and execution for each experiment were identical, with the following exceptions: (1) copper, a component of the production medium, was inadvertently supplemented at one‐tenth of the standard concentration in Experiment 3 for mAb B cell lines; and (2) mAb C cell line #6 yielded very low product titer (<0.05 g/L) using both cryopreserved and non‐cryopreserved cells in Experiment 1 and was therefore not carried forward for subsequent production culture assays (i.e., Experiments 2 and 3). To prepare for a production culture assay, cells from seed train cultures were centrifuged (830g, 5 min) and cell pellets were resuspended in proprietary chemically defined production medium without MSX. Production cultures were inoculated at 80 mL working volume in 250 mL vent‐capped Erlenmeyer flasks (Corning) at a VCC of ∼1 × 106 cells/mL. These cultures were agitated by an orbital shaker (150 rpm, 25 mm throw) in a humidified incubator with a 5% CO2 overlay and maintained at 37°C for the first three days, and at 35°C thereafter until the cultures were harvested on day 14. On days 3, 7, and 10 post inoculation, production cultures were supplemented with a chemically defined nutrient feed at 10% (v/v).

Culture samples taken on days 3, 7, 10, and 14 were analyzed immediately for VCC and viability using the Vi‐Cell XR (Beckman Coulter). Day 14 supernatants were stored frozen at −80°C until they were analyzed for mAb titer and product quality (i.e., size, charge, and glycosylation distributions) as previously described.[16] Product quality metrics were not provided when the mAb titer was low (<0.3 g/L) because of insufficient material for testing in the high‐throughput assays.

Flow cytometric analysis was performed on all cell lines using seed train cultures originating from both cryopreserved and non‐cryopreserved cells within a week of initiating Experiments 1, 2, and 3 (Table ). The mAbs expressed by the cell lines used for studying the impact of cryopreservation (i.e., mAbs A, B, and C) belong to the IgG1 subclass. Likewise, the mAbs expressed by the cell lines used for evaluating the flow cytometric analysis method (i.e., mAbs D–M) also belong to the IgG1 subclass and were generated as previously described.[15]

Cells used for flow cytometric analysis were taken from seed train cultures 3–4 days after subculturing (∼4 × 106 cells, >90% viability), unless stated otherwise. The media used for seed train and production cultures were chemically defined but differed in composition. For example, the seed train medium contained 25 μM MSX to maintain selective pressure, whereas the production medium did not contain MSX. After centrifugation (830g, 5 min), cell pellets were rinsed with phosphate‐buffered saline (PBS) prior to cell fixation and permeabilization using BD Cytofix/Cytoperm solution kit (BD Biosciences Cat #554714) according to the manufacturer's instructions.

To perform intracellular mAb staining, a 1.4 mg/mL stock solution of goat anti‐human IgG (H + L) F(ab′)2 conjugated to FITC (Jackson ImmunoResearch Laboratories, Inc. Cat #109‐096‐088) was diluted 50‐fold in 1× wash buffer and added to the cell pellets. After incubating these samples at room temperature (25 min) with gentle agitation, the cells were washed and resuspended in 1× wash buffer.

To perform dual staining of dead cells and intracellular mAb, a blue fluorescent reactive dye for detecting dead cells was used in conjunction with the FITC‐conjugated goat anti‐human IgG (H + L) F(ab′)2 described above for detecting intracellular mAb. The blue dye, obtained from the LIVE/DEAD Fixable Dead Cell Stain Kit (Thermo Fisher Scientific, Cat #L23105), was used according to the manufacturer's instructions. In brief, each cell pellet resuspended in PBS was mixed with equal volume of reconstituted blue dye and incubated with light protection (room temperature, 30 min). After washing the cells with PBS, intracellular mAb staining was performed as described above.

To perform dual staining of DNA and intracellular mAb, the propidium iodide (PI) fluorescent dye (Thermo Fisher Scientific) for detecting DNA was used in conjunction with the FITC‐conjugated goat anti‐human IgG (H + L) F(ab′)2 described above. This dual staining was performed based on a previously published method.[17] In brief, intracellular mAb staining was first performed as described above. After the final PBS wash, the cells were resuspended in RNase A solution (0.2 mg/mL; Sigma‐Aldrich) and incubated (room temperature, 30 min). DNA was then stained by adding PI solution (50 μg/mL) and incubating the cell suspension in the dark (2–8°C, 45 min).

To perform flow cytometric analysis of mixed population of cells, the appropriate volumes from the two seed train sources were mixed to achieve the targeted ratio of the two cell populations. The mixed cells were then fixed, permeabilized, and stained as described above using FITC‐conjugated goat anti‐human IgG (H + L) F(ab′)2.

All final cell suspensions were passed into round‐bottomed polystyrene tubes (BD Biosciences Cat #352235) through cell strainers (35 µm mesh size). Flow cytometry was performed on a BD FACSCalibur instrument (BD Biosciences) with 15 mW, 488 nm argon ion laser, 530 ± 30 nm bandpass filter, and gating set for 20,000 events per sample. Dead cells, debris, and clumps were gated out based on forward scatter and side scatter. The parental (i.e., non‐transfected) CHO cells were used as the negative control. Flow cytometry for intracellular mAb detection was based on FITC fluorescence. Flow cytometry to differentiate live and dead cells was performed using excitation wavelength of 350 nm and emission wavelength of 450 nm. Flow cytometry to assess PI fluorescence was performed using a 585 ± 42 nm bandpass filter; the PI fluorescence signal was used to quantify the cellular DNA content. FITC analysis was performed on the FL1 channel with a log scale and PI analysis was conducted on the FL3 channel with a linear scale. Intracellular mAb staining (based on FITC fluorescence) and cell cycle distribution (based on PI fluorescence) were analyzed using the FlowJo flow cytometry analysis software (FlowJo, LLC).

Integrated viable cell concentration (IVCC), an indicator of overall cell growth, was calculated using the multiple‐application trapezoidal rule[18] :where i = sample number (in which samples 1, 2, 3, 4, and 5 were collected on days 0, 3, 7, 10, and 14, respectively), ti = day of sample i collection, and VCCi = viable cell concentration for sample i.

Cell‐specific productivity (Qp) for each cell line was calculated in units of pg cell−1 day−1 (pcd). Qp was measured using the slope of the linear fit obtained from plotting mAb titer vs. IVCC in the production cultures based on the relationship:

Statistical tests were performed using JMP 11.1.1 software (SAS Institute Inc.) to search for significant effects of cryopreservation on cell culture performance and product quality. Specifically, Student's t‐tests were conducted to assess for statistically significant differences (P < 0.05) between cryopreserved and non‐cryopreserved cells based on their performances in seed train and production cultures. For the seed train cultures, end of passage viability and growth rate in seed train cultures used to source production culture assays were analyzed. The growth rate was calculated by taking the natural log of the ratio of VCC at the end of passage to the VCC at the start of the passage, and dividing this number by the passage duration:

The population doubling time was calculated by dividing the natural log of 2 by the growth rate:

The number of population doublings, referred to herein as population doubling level (PDL), was calculated by dividing the culture duration (i.e., cumulative time in seed train passaging) by the population doubling time:

For the PDL calculations performed in this work, the PDL value of zero was defined as the time of cryopreservation.

To assess the recovery from thaw for cryopreserved mAb A, B, and C cell lines, Student's t‐tests were applied to determine statistically significant differences between seed train passages. For the mAb A, B, and C production cultures, viability, IVCC, mAb titer, Qp, as well as mAb size, charge, and glycosylation distributions were analyzed. For these cell culture and product quality attributes, Student's t‐tests were used to assess for statistically significant differences between cryopreserved and non‐cryopreserved cells in the following ways: (1) all cell lines across all experiments; (2) mAb A cell lines across all experiments; (3) mAb B cell lines across all experiments; (4) mAb C cell lines across all experiments; (5) all cell lines within Experiment 1; (6) all cell lines within Experiment 2; and (7) all cell lines within Experiment 3. Only the lowest P value (out of the seven P values obtained by Student's t‐tests) is denoted for each attribute in the text herein. All P values are provided in the Supporting Information Tables S1–S6.

To assess the more immediate effects of the freeze‐thaw process (i.e., cryopreservation and recovery from thaw), we monitored the performance of cryopreserved recombinant CHO cell lines (expressing mAb A, B, or C) in the initial seed train passages after thaw. Of the 24 cell lines that had undergone the freeze‐thaw process, four mAb C cell lines showed <80% viability at the end of the thaw passage (Figure A). These four cell lines also showed the lowest growth rates (<0.2 day−1) in the thaw passage (Figure B). Two of these cell lines (mAb C cell lines #1 and #6) improved in viability (>90%) and growth rate (>0.4 day−1) by the second passage after thaw; the other two cell lines (mAb C cell lines #7 and #8) subsequently recovered in viability (>90%) and growth rate (>0.4 day−1) by the fourth passage after thaw. All eight mAb C cryopreserved cell lines showed no statistically significant difference in growth rates (P > 0.44; Supporting Information Table S1) between the fourth passage after thaw and the passage used to source the first production culture assay (Experiment 1). Although the Student's t‐test showed a statistically significant difference in viability (P < 0.05; Supporting Information Table S1) between these two non‐consecutive seed train passages for the mAb C cell lines (spaced ∼1 month apart), the actual difference was small (∼1%) and within the variability associated with culture viability measurements by the Vi‐Cell XR instrument.[19]

By the start of the first production culture assay (Experiment 1) ∼1 month after recovery from thaw, the seed train viabilities were similarly high (>96%) for all 24 cryopreserved cell lines and their 24 non‐cryopreserved counterparts (Figure A). The growth rates for the 48 seed train cultures used to source Experiment 1 were also similar between the cryopreserved cell lines and their non‐cryopreserved counterparts (Figure B). At the start of the second and third production culture assays (Experiments 2 and 3), initiated ∼1 and ∼2 months after Experiment 1, the viabilities remained high (>96%) and the growth rates remained similar between seed trains originating from cryopreserved and non‐cryopreserved cell lines (Figure ).

Between the cryopreserved and non‐cryopreserved cell lines, there were no statistically significant differences in viability (P > 0.30) and growth rate (P > 0.18) for the seed train cultures used to source Experiments 1, 2, and 3 when the data were analyzed in the following ways (Supporting Information Table S2): (1) all cell lines across all experiments (i.e., using all available data); (2) cell lines for a specific mAb (i.e., mAb A, mAb B, or mAb C) across all experiments; and (3) all cell lines within a specific experiment (i.e., Experiment 1, 2, or 3). The analogous set of comparisons was used to assess cryopreserved vs. non‐cryopreserved cells in studies described hereinafter.

To assess the longer‐term effects of the freeze‐thaw process, we monitored the performance of cryopreserved recombinant CHO cell lines and their non‐cryopreserved counterparts in a series of three production culture assays spaced ∼1 month apart (Experiments 1, 2, and 3). Within each cell line expressing a specific mAb (i.e., mAb A, B, or C), culture viability (Figure A) and cumulative cell growth, as represented by IVCC (Figure B), were similar at the time of harvest (day 14) for a given experiment whether cryopreserved or non‐cryopreserved cells were used. For example, final viability was similarly low for all eight mAb B cell lines in Experiment 3, irrespective of cryopreservation.

Likewise, productivity was similar in terms of mAb titer (Figure A) and Qp (Figure B) for a given experiment whether cryopreserved or non‐cryopreserved cells were used. For example, productivity values were similarly low, regardless of cryopreservation, for the cell lines with low product titer (<0.3 g/L mAb) and low Qp (<3 pcd).

Lower final viability, IVCC, and titer (Figures and ) were observed for all mAb B cell lines in Experiment 3 compared with Experiments 1 and 2. This consistently poorer culture performance in Experiment 3 is attributed to the unintended 10‐fold lower copper concentration in the production medium (which only occurred in Experiment 3 for the 16 mAb B cultures). These results are consistent with our previous findings on the negative impact of lower copper concentration on these cell culture attributes, as well as on lactate metabolism.[20] Final lactate concentrations were consistently higher for all 16 mAb B cultures in Experiment 3 than in Experiments 1 and 2 (data not shown), which likely impacted cell viability. This negative impact to culture performance was independent of whether the cells were previously cryopreserved.

Although all eight mAb C cell lines yielded product titers >1 g/L during clone screening conducted prior to cryopreservation (data not shown), six of these cell lines subsequently yielded titers <1 g/L in Experiment 1. In particular, mAb C cell line #6 lost productivity by Experiment 1; both cryopreserved and non‐cryopreserved cells yielded <0.05 g/L of mAb C (with Qp <0.5 pcd) and there was insufficient mAb for product quality assays. Therefore, both the cryopreserved and non‐cryopreserved seed train cultures for mAb C cell line #6 were discontinued after Experiment 1. Although mAb C cell line #7 also showed low productivity (∼0.5 g/L mAb C and Qp ∼3 pcd) in Experiment 1, this productivity was adequate for supporting product quality assays. Hence the parallel seed trains for mAb C cell line #7 were maintained. There were no statistically significant differences in final viability (P > 0.17), final IVCC (P > 0.64), mAb titer (P > 0.70), and Qp (P > 0.75) between the production cultures sourced from cryopreserved and non‐cryopreserved cells, when the data were analyzed as described earlier for seed train performance (Supporting Information Table S3).

To assess the impact, if any, of the freeze‐thaw process on mAb product quality, we measured the typical quality attributes for recombinant mAb products—size, charge, and glycosylation distributions—generated from cells sourced from cryopreserved and non‐cryopreserved seed trains.

Within each cell line expressing a specific mAb, size distribution for the mAb product was generally similar in terms of monomer (Figure A), high molecular weight species (Figure B), and low molecular weight species (data not shown) for a given experiment whether cryopreserved or non‐cryopreserved cells were used. The size distributions for cell lines in some experiments were not provided because of insufficient product (mAb titer <0.3 g/L; Figure A) for reliable analysis by the in‐house high‐throughput platform version of the size exclusion chromatography assay. Although some differences were observed in the high molecular weight species for mAb C, there was no consistent pattern with respect to cryopreservation and the differences were not statistically significant (P > 0.92). There were no statistically significant differences in monomer (P > 0.36), high molecular weight species (P > 0.69), or low molecular weight species (P > 0.24) between the cryopreserved and non‐cryopreserved cells, when the data were analyzed as described earlier for seed train performance (Supporting Information Table S4).

Within each cell line expressing a specific mAb, the charge distribution for the product was similar in terms of acidic species (Figure A), basic species (Figure B), and main species (data not shown) for a given experiment whether cryopreserved or non‐cryopreserved cells were used. The charge distributions for cell lines in some experiments were not provided because of insufficient product (mAb titer <0.3 g/L; Figure A) for reliable analysis by the high‐throughput imaged capillary isoelectric focusing assay. There were no statistically significant differences in acidic species (P > 0.31), basic species (P > 0.21), or main species (P > 0.37) between the cryopreserved and non‐cryopreserved cell lines, when the data were analyzed as described earlier for seed train performance (Supporting Information Table S5).

Within each cell line expressing a specific mAb, glycosylation distribution for the mAb product was similar in terms of G0 species (Figure A), Man5 species (Figure B), and G0‐F species (data not shown) for a given experiment whether cryopreserved or non‐cryopreserved cells were used. The structures of these N‐linked glycosylated species are detailed in Gennaro and Salas‐Solano.[21] The glycosylation distributions for cell lines in some experiments were not provided because of insufficient product (mAb titer <0.3 g/L; Figure A) for reliable analysis by the high‐throughput capillary electrophoresis assay. There were no statistically significant differences in G0 species (P > 0.54), Man5 species (P > 0.66), or G0‐F species (P > 0.45) between the cryopreserved and non‐cryopreserved cell lines, when the data were analyzed as described earlier for seed train performance (Supporting Information Table S6).

To assess the impact, if any, of the freeze‐thaw process on other aspects of mAb expression that may not be revealed by standard production culture assays, we used flow cytometry to investigate the heterogeneity in intracellular mAb expression. Cells sourced from cryopreserved and non‐cryopreserved seed train cultures were fixed, permeabilized, and then stained with FITC conjugated anti‐IgG, similar to that described by Dorai et al.[10] The cultures were assessed by flow cytometry for one of two categorical outcomes on fluorescence intensity histograms: (1) single peak, or (2) non‐single peaks. A homogeneous population of cells with relatively uniform mAb production is expected to produce a single peak in intracellular mAb expression. By contrast, a heterogeneous population of high and low mAb producers is expected to produce double peaks in which the low producers would be represented by the less fluorescent (i.e., leftmost) peak.[10] [12] [13] [14]

The study design facilitated pair‐wise comparison of flow cytometric profiles for each cryopreserved cell line and its non‐cryopreserved counterpart within each experiment (Figure ). The pair‐wise comparisons did not show any noticeable effects of the freeze‐thaw process on intracellular mAb expression for the recombinant CHO cell lines studied here. For each cell line expressing a specific mAb, flow cytometric profiles were similar within a given experiment whether cryopreserved or non‐cryopreserved cells were used. Specifically, if the cryopreserved cells showed a single peak by flow cytometry, the corresponding non‐cryopreserved cells also showed a single peak. Conversely, if the cryopreserved cells showed non‐single peaks, the corresponding non‐cryopreserved cells also showed non‐single peaks. In one example, mAb A cell line #2 showed a single peak at the start of Experiment 1 and non‐single peaks at the start of Experiment 3, regardless of whether the flow cytometry was performed on cryopreserved (Figure A) or non‐cryopreserved (Figure B) cells. In another example, mAb B cell line #2 showed a single peak at the start of Experiment 1 and non‐single peaks by the start of Experiments 2 and 3, regardless of whether the flow cytometry was performed on cryopreserved (Figure C) or non‐cryopreserved (Figure D) cells.

All 24 cell lines displayed single peaks before cryopreservation (data not shown), which was a clone selection criterion. However, by the start of Experiment 1, ∼1 month after cryopreservation, eight cell lines showed non‐single peaks (data not shown), indicating increased heterogeneity in terms of intracellular mAb expression. Twelve cell lines showed distinct changes in flow cytometric profiles between Experiments 1 and 3 such that non‐single peaks were observed within at least one experiment (Table ). The other 12 cell lines consistently showed single peaks (for both cryopreserved and non‐cryopreserved cells across Experiments 1, 2, and 3). Interestingly, none of the mAb C cell lines fell into the category of cell lines with consistent single peaks.

Summary of Flow Cytometry and Qp Profiles for 24 Cell Lines

• 4 Flow cytometry analysis for each cell line was performed on both cryopreserved and non‐cryopreserved seed train cultures used to source Experiments 1, 2, and 3. Flow cytometric profiles (in terms of fluorescence intensity histograms) were assessed for the presence of a consistent single peak across Experiments 1, 2, and 3 (spanning ∼2–3 months of seed train passaging).
• 5 Analysis of cell age impact on Qp for each cell line was performed by calculating the percent decrease in Qp between Experiment 1 and Experiment 3.
• 6 Only one cell line showed difference in outcome for Qp decrease criterion when comparing cells sourced from cryopreserved vs. non‐cryopreserved seed train cultures. Qp decrease for mAb C cell line #2 was ∼16% and ∼24% using cryopreserved and non‐cryopreserved cells, respectively. The more conservative outcome (Qp decrease ≥20%) was applied for this cell line.

To assess the potential impact of sample day, cell viability and cell cycle distribution on intracellular mAb expression, additional experiments were conducted on other recombinant CHO cell lines. When three CHO cell lines expressing mAb D were stained for intracellular mAb, similar histograms were obtained for each cell line taken from days 1 and 4 of seed train cultures (Supporting Information Figure S1). When four CHO cell lines (expressing mAbs E, F, G, and H, respectively) were dual stained for dead cells and intracellular mAb, similar histograms were also obtained for each cell line, with and without gating for live cells (Supporting Information Figure S2). When six CHO cell lines (expressing mAbs E, G, I, J, K, and L, respectively) were dual stained for DNA and intracellular mAb, similar histograms were again obtained for each cell line, with and without gating for cells in the G0/G1 phase of the cell cycle (Supporting Information Figure S3).

To assess the ability of this flow cytometric analysis to distinguish between mixed populations of cells, different CHO cell lines were mixed together and immediately fixed, permeabilized, and stained with FITC conjugated anti‐IgG. In the first study on a mixed population, the recombinant CHO cell line expressing mAb G was spiked with three different levels (2, 5, and 10%) of the non‐transfected CHO parental host. The less fluorescent (i.e., leftmost) secondary peak increased in height with increasing levels of non‐transfected parental host (Supporting Information Figure S4). The difference in mean fluorescence intensity (MFI) for the two peaks was between 1 and 2 log. In the second study on mixed populations, the mAb G cell line (with Qp ∼11 pcd) was spiked with a recombinant CHO cell line expressing mAb M (with Qp ∼ 20 pcd) at seven different levels (10, 20, 40, 50, 60, 80, and 90%). mAb G and mAb M cells in isolation, as well as when mixed at the seven different combinations, displayed single peaks by flow cytometric analysis (Supporting Information Figure S5). In subsequent studies on mixed populations using cell lines with single flow cytometric peaks, two cell lines expressing the same or different mAbs were mixed in equal proportions. These pairs of cell lines had similar Qp or Qp that differed by several fold (e.g., ∼2 and ∼16 pcd; ∼8 and ∼23 pcd). In all instances, the resulting fluorescence intensity histograms from flow cytometric analysis of the mixed cells showed a single peak (data not shown).

In our study design, the experiment progression was confounded with cell age because each successive experiment was performed at an older cell age, as represented by time in culture (Table ). The only clear trend observed across the experiments in terms of product quality pertained to mAb B in Experiment 3 (Figure ). Higher levels of acidic species and lower levels of basic species were observed in the charge distributions for all mAb B cultures in Experiment 3 than in Experiments 1 and 2. This consistent trend may be attributed to the unintended 10‐fold lower copper concentration in the production medium (which only occurred in Experiment 3 for the 16 mAb B cultures). These results are consistent with our previously reported observations showing lower levels of basic species with lower initial levels of copper.[20] More importantly, the impact to charge distribution in Experiment 3 for mAb B cell lines was independent of cryopreservation.

Other recombinant CHO cell lines have shown changes within a slightly shorter time frame in terms of intracellular IgG expression by flow cytometry and product titer.[7] [10] In particular, Dorai et al.[10] showed a correlation between the size of the secondary (i.e., less fluorescent) peak and the percent decline in titer. A recombinant CHO cell line could yield a stable product titer with cell age despite a time‐dependent decrease in Qp because of the concomitant increase in cell growth and hence IVCC (Eq. ). Such a cell line may be deemed unstable in productivity on a per cell basis because of the declining Qp with cell age. Therefore, we selected Qp instead of mAb titer as the surveillance marker for production stability. Given that our intended purpose for flow cytometric analysis was to eliminate cell lines with potentially unstable productivity, quantitative assessments (for ranking individual clones) was not required. In addition, intracellular MFI was difficult to reproduce across assay sessions for the same cell line and did not necessarily correlate with titer or Qp (data not shown). Therefore, unlike Dorai et al.,[10] we opted to evaluate the flow cytometric data in a qualitative instead of a quantitative manner. As described earlier in the flow cytometry section (within Results), we focused on the emergence of additional, lower‐expressing peak(s) with cell age as a potential indicator for Qp decline.

To investigate the potential connection between flow cytometric profiles and Qp, we compared changes in these outputs between Experiments 1 and 3 for these 24 CHO cell lines. We assumed that a Qp decrease of <20% could be accounted for by variability in IVCC and mAb titer measurements. Therefore, for a cell line to be deemed unstable in Qp, we applied the criterion of Qp decrease by ≥ 20%.

Based on this Qp criterion, only one of 24 cell lines showed a difference in the assessment outcome as a result of cryopreservation status. Specifically, for mAb C cell line #2, the Qp decrease in the production cultures was ∼16% using cryopreserved cells and ∼24% using non‐cryopreserved cells. For this cell line, we applied the more conservative outcome (i.e., Qp decrease of ≥20%). By our qualitative assessments for this panel of 24 CHO cell lines, the lack of a consistent single peak in flow cytometric profiles over cell age was indicative of Qp decline. All 12 cell lines with inconsistent flow cytometric profiles also showed ≥20% decrease in Qp between Experiments 1 and 3, including all eight mAb C cell lines.

However, the converse did not hold; the presence of a consistent single peak in flow cytometric profiles was not always indicative of Qp stability. For the 12 cell lines with consistent flow cytometric profiles (i.e., single peak), five showed ≥20% decrease in Qp between Experiments 1 and 3 (Table ). All five cell lines (mAb A cell lines #3 and #8, and mAb B cell lines #1, #7, and #8) showed single peaks with similar shape and MFI across Experiments 1, 2, 3 using cryopreserved and non‐cryopreserved cells (data not shown).

Cryopreservation did not exert any notable impact on product quality, flow cytometric profiles or Qp decline with cell age. For example, the 24 cryopreserved cell lines showed the same flow cytometric profiles as their 24 non‐cryopreserved counterparts at a similar cell age (i.e., at the time of the same experiment). This enabled us to summarize the outcomes from flow cytometric analyses in Table without having to differentiate between cryopreserved and non‐cryopreserved cells.

Beyond the initial passages after thaw (Figure ), the freeze‐thaw process associated with cryopreservation and recovery from thaw did not have any subsequent effects on the performance of the 24 recombinant CHO cell lines studied here. Specifically, cell culture performance and product quality—based on three sets of production culture assays (Experiments 1, 2, and 3) spanning ∼2–3 months in seed train passaging—were similar between the 24 cryopreserved cell lines and their corresponding non‐cryopreserved counterparts (Figures ).

Our findings corroborate earlier observations that cryopreservation did not have an effect on viable cell density, viability, and recombinant protein titers for murine myeloma (NS0) cultures.[2] To augment the groundwork established by Barnes et al.,[2] our study involved 24 recombinant CHO cell lines that were cryopreserved in serum‐free chemically defined media. In addition, we characterized the three mAbs (in terms of size, charge, and glycosylation distributions) to confirm appropriate and consistent product quality, which is a requirement for robust biopharmaceutical manufacturing.[3] Taken together, both studies support the use of cryopreservation to create mammalian cell banks.

To investigate the underlying mechanisms for Qp instability observed in CHO cell lines expressing a specific mAb, Misaghi et al.[8] compared CHO cultures sourced from cryopreserved cells with their non‐cryopreserved counterparts and found higher intracellular accumulation of mAb in the former. Although this finding was based on Western blot analysis, such differences in intracellular mAb accumulation may be detected by the flow cytometric approach described here (i.e., using FITC conjugated anti‐IgG). Hence we analyzed the effect of cryopreservation on intracellular product expression by flow cytometry. This additional characterization—performed on cryopreserved and non‐cryopreserved seed train cultures before initiating Experiments 1, 2, and 3showed a lack of impact from cryopreservation. To our knowledge, this high‐throughput flow cytometry assay has not been previously used to characterize the potential long‐term effect of freeze‐thaw on intracellular mAb expression.

The only observations attributed to the potential stress and damage from the freeze‐thaw process were found in the initial post‐thaw passages for four out of 24 cell lines (Figure ). Despite the poor performance (in terms of end of passage viability and growth rate) out of thaw, these four cell lines recovered by the fourth passage after thaw to yield similar viability and growth rates as the other twenty cell lines. More importantly, freeze‐thaw exerted no lasting impact. Specifically, the flow cytometric and product quality profiles as well as the subsequent decrease in titer and Qp for these four cell lines were similar for cultures sourced from cryopreserved and non‐cryopreserved cells. Therefore, there is no evidence to suggest that initial post‐thaw performance challenges lead to the outgrowth of a subpopulation, in spite of the dynamic nature of CHO cells.[3] [22] [23]

Although the flow cytometric analysis showed that freeze‐thaw alone did not lead to notable changes in intracellular mAb expression, it showed the effects of cell age: 12 of the 24 cell lines displayed altered flow cytometric profiles (i.e., increased intracellular mAb expression heterogeneity) over ∼2–3 months of passaging in seed train cultures, with concurrent Qp decline of ≥20% (Table ). Dorai et al.[10] proposed using a similar flow cytometric method to identify and eliminate unstable clones because of their observed correlation between decrease in mAb titer and appearance of a secondary population of cells with lower intracellular mAb content. Our findings substantiate their proposal and justify the preferential selection of cell lines with sustained single peaks during clone screening.

Although our findings demonstrate the utility of this flow cytometric approach as a checkpoint during clone screening, they also highlight its limitations—this checkpoint for Qp stability is susceptible to false negatives. For example, five of the 24 cell lines presented consistent single peaks (with similar shape and MFI) by flow cytometry, but these five cell lines did not satisfy our Qp stability criterion (<20% Qp decrease between the first and last production culture assays, ∼2–3 months apart). Moreover, <30% of the cell lines tested met our Qp stability criterion (Table ).

To investigate the potential causes for these false negative readings (i.e., cell lines with Qp decrease ≥20% but showed consistent single peaks by flow cytometry), factors that may affect the flow cytometric profiles were further evaluated. Cells taken from days 1 and 4 of seed train cultures for three mAb D cell lines yielded similar histograms (Supporting Information Figure S1). Therefore, the use of cells taken from day 3 or day 4 of a seed train culture, which was our standard procedure, should not affect the flow cytometry outcome. Likewise, hybridoma cells taken from both day 3 and day 5 cultures yielded similarly shaped histograms with double peaks when analyzed by flow cytometry for intracellular antibody expression.[12]

To explore the impact of cell viability on flow cytometric profiles, we performed dual staining of dead cells and intracellular mAb. Gating for live cells had a negligible impact on the resulting fluorescence intensity histograms (Supporting Information Figure S2), thereby demonstrating that culture viability did not contribute to false negatives. Although Dorai et al.[10] found a positive correlation between production instability and apoptotic markers (caspase 3, R = 0.85; Annexin V, R = 0.59), they subsequently noted that further assessment of caspase 3 expression across a large number of cell lines showed a significant degree of variability. They concluded that caspase 3, the most promising apoptotic marker that they identified, could only be used in a qualitative manner to indicate potential cell line instability.

To explore the impact of cell cycle on flow cytometric profiles, we performed dual staining of DNA and intracellular mAb. Cells in the G0/G1 phase of the cell cycle should contain less DNA than the cells in the S and G2/M phases of the cell cycle. G0/G1‐gated and non‐gated cells showed similar histograms (Supporting Information Figure S3). This observation is not surprising because fluorescence intensity on the histograms is expressed on a log scale; the width of the base for a single peak can extend beyond a 10‐fold (i.e., 1 log) range in fluorescence intensity, as observed for intracellular IgG expression in recombinant CHO cells (Figure , Supporting Information Figures S1–S5)[10] and hybridoma cells.[14] [17] In addition, the double peaks observed in recombinant CHO cells (Figure and Supporting Information Figure S3) typically displayed a peak‐to‐peak separation in fluorescence intensity of 1–2 log.[10] Attempts to increase productivity of CHO cells by cell cycle control have yielded up to several‐fold increase in Qp.[24] [25] [26] If CHO cells in different phases of the cell cycle differed in intracellular mAb expression by several fold, our flow cytometric analysis is unlikely to have the level of sensitivity to detect such differences in intracellular mAb expression.

To explore the sensitivity of our flow cytometric approach in detecting mixed populations of cells, we performed multiple mixing studies using different CHO cell lines. With the exception of spiking the mAb G cell line with non‐transfected CHO host cells (Supporting Information Figure S4), none of the other mixed population studies showed non‐single peaks on the fluorescence intensity histograms (Supporting Information Figure S5), even for cell lines with several‐fold differences in Qp. These mixing studies confirm the limited sensitivity of the flow cytometric analysis.

Factors beyond intracellular mAb content can affect the Qp of a cell. Such factors include intracellular mAb degradation as well as mAb secretion capability (and potential bottlenecks in the secretory pathway),[2] which cannot be directly assessed by our flow cytometric approach. Taken together, the observed outcomes and the limited sensitivity of the flow cytometric approach underscore the continued need for standard production culture assays to screen for stable cell lines.

It may not be a coincidence that only mAb C cell lines displayed poor recovery from thaw (Figure ) and all eight mAb C cell lines showed Qp decrease of ≥20% (Table ). Prior studies have indicated that mAb C is difficult to express (data not shown). To cover a more diverse range of mAb products and their associated challenges during cell line development, process development, and cell banking, we included mAb C as one of the three products tested in this study.

In the study by Misaghi et al.,[8] CHO cell lines transfected to constitutively express a specific mAb exhibited product‐induced instability that was exacerbated by freeze‐thaw. The apparent contradiction between our conclusion here that freeze‐thaw did not exacerbate cell line instability and observations by Misaghi et al. may be reconciled by the differences in mAb. Some mAb A and B cell lines were able to meet our Qp stability criterion. Although mAb C is considered a difficult‐to‐express molecule, mAb C cell line #2 displayed only borderline Qp instability (Qp decrease over ∼3 months was ∼16% and ∼24% for cryopreserved and non‐cryopreserved cells, respectively). By contrast, the specific mAb tested by Misaghi et al. consistently induced substantial Qp loss in recombinant cell lines generated through different cell line development approaches; that specific mAb may be an unusual exception.

Only seven out of 24 recombinant CHO cell lines (four mAb A and three mAb B cell lines) showed Qp decrease of <20% between Experiments 1 and 3 (spanning ∼2–3 months of passaging), despite maintaining seed train cultures under MSX selective pressure (Table ). Even upon excluding all the mAb C cell lines from the Qp analysis, the percentage of cell lines with Qp decrease of ≥20% remained high (>50%). This percentage is in line with the findings by Kim et al.[5] in which five out of 10 mAb‐producing CHO‐K1 cell lines showed Qp decrease of >20% after ∼41–81 generations.

While many studies have demonstrated the production instability of recombinant CHO cells (as reviewed extensively by Barnes et al.[11] and discussed recently by Wurm and Wurm[22] ), the bulk of these studies focused on CHO cells utilizing the DHFR‐based MTX selection system. Of the few studies that focused on CHO‐K1 cells utilizing the GS‐based MSX selection system,[5] [7] [9] [10] only the study by Kim et al.[5] lends to an appropriate comparison to our study here. This is because of the fundamental commonalities in both studies: (1) GS‐mediated gene amplification using MSX concentration beyond 50 μM was not applied during cell line development; (2) seed train cultures were maintained under MSX selective pressure; (3) multiple cell lines expressing different mAbs were studied; and (4) Qp decrease over a similar timeframe was quantifiable based on fed‐batch production culture assays. In the study by Bailey et al.,[9] only one recombinant CHO cell line was investigated for Qp decrease (>30% between passages 40 and 60). In the study by Dorai et al.,[10] the majority of the recombinant CHO cell lines tested were maintained without MSX selection, and stability was quantified by titer, not Qp. Of the several cell lines maintained in the presence of MSX, only one was carried past the fifth passage, and it showed a 3% titer decrease by the eleventh passage.[10] In the study by Jun et al.,[7] 10 amplified CHO cell lines expressing the same antibody were maintained at 100, 500, and 1000 μM MSX in static plate cultures; based on analysis of seed train cultures, only one cell line showed Qp decrease of <20% by the twelfth passage, whereas the other nine cell lines showed average Qp decrease of >60%.

Although the underlying molecular mechanisms for the ≥20% Qp decline over ∼2–3 months of passaging (observed here for 17 out of 24 recombinant CHO cell lines) are not related to cryopreservation, other mechanisms uncovered by prior research may not be excluded. Some of these mechanisms include loss of transgene copies and/or transcriptional silencing,[5] [7] [11] [27] [28] [29] [30] limitations in antibody secretion[8] [11] [31] and changes in metabolism.[9]

These findings conclusively rule out freeze‐thaw associated stress and damage as the primary cause for Qp decline in 17 of the 24 recombinant CHO cell lines tested here. Therefore, the previous observation for a different mAb, in which freeze‐thaw exacerbated product‐induced cell line instability,[8] may be considered an exception to the norm. The other previous observations on productivity loss after freeze‐thaw did not include controls to isolate the effects of cell age and cryopreservation.[6] [7] Since both factors were confounded in those studies, the observed productivity loss may be a consequence of cell age, instead of cryopreservation. Together with the evaluation of cryopreservation using NS0 cell lines,[2] our findings support the continued use of cryopreservation to generate mammalian cell banks, which provide the foundation for research, development, and manufacturing operations.

To build upon the groundwork established by Dorai et al.,[10] we further illustrated the utility of flow cytometry in monitoring intracellular mAb expression as a supplementary tool to eliminate unstable cell lines based on the presence of non‐single peaks. However, standard production culture assays still remain essential for clone selection since the flow cytometric approach could not identify Qp instability in 5 of 17 cell lines which consistently exhibited a single peak (Table ); the presence of a single peak by flow cytometric analysis is a necessary but not sufficient condition for identifying Qp stability (<20% decline after ∼2–3 months of seed train passaging).

Qp stability is an important criterion but it is not the only factor considered during clone selection. Typically, only cell lines that produce material with acceptable product quality attributes are selected during clone sreening.[3] Unlike standard production culture assays—which generate harvests for product quality analyses—our flow cytometry assay uses cells from seed train cultures and does not provide product quality outcomes. Therefore, this flow cytometric approach lacks the product quality assessments required for clone selection. Both the potential and limitations of the flow cytometric approach, as demonstrated in our study, support our current clone screening workflow in which flow cytometry functions as a checkpoint to narrow down the number of candidate cell lines. Therefore, flow cytometric outcomes enable us to focus the more labor‐intensive production culture assays and cell banking operations on cell lines that passed the flow cytometric screen.

Cell line instability remains an ongoing challenge for the CHO culture‐based biopharmaceutical industry, as illustrated by its prevalence within our study of 24 recombinant CHO cell lines expressing three distinct mAbs, and by numerous earlier studies.[5] [7] [9] [11] [27] [28] [29] [30] These findings are not surprising because the inherent fluidity of the CHO genome increases its susceptibility to instability.[3] [22] [23] However, it may be possible to overcome these innate challenges without relying on extensive clone screening. For example, with the recent advances in site‐specific integration, we may capitalize on the cellular genetic plasticity of CHO cells to select or engineer a host with desired traits[5] [23] [32] and also locate a targeted integration site that maintains stable and high‐level expression of the transgene.[33] [34] [35] Regardless of the future strategies to enhance cell line stability, we can bank on cryopreservation to lay the foundation for these advances.

The findings here demonstrate that the routine cell banking process and the potential stress and damage during cryopreservation and recovery from thaw had no lasting effects on recombinant CHO cell lines. Across the panel of 24 recombinant GS/MSX‐CHO cell lines which were transfected from a CHO‐K1 host to express one of three mAbs, no significant differences were observed as a result of freeze‐thaw beyond the initial post‐thaw passages. Specifically, cultures sourced from cryopreserved cells and their non‐cryopreserved counterparts yielded similar performance (in terms of growth, viability, and productivity), product quality (in terms of size, charge, and glycosylation distributions), and flow cytometric profiles (in terms of intracellular mAb expression).

The findings here also testify to the prevalence of Qp decline in recombinant CHO cell lines developed using the GS‐based MSX selection system. Seventeen of the 24 recombinant GS/MSX‐CHO cell lines tested here showed Qp decline over time (≥20% Qp decrease after ∼2–3 months of passaging) despite maintaining MSX selective pressure in the seed train cultures. These data augment the relatively scarce statistics on cell line stability for recombinant CHO cells derived from the GS‐based MSX selection system, in contrast to the well‐studied DHFR‐based MTX selection system. These findings also illustrate the utility of flow cytometry as a high‐throughput tool to eliminate cell lines with heterogeneous intracellular mAb expression. However, our flow cytometric assay is only suitable as a supplementary tool to the traditional production culture assays since it could not identify Qp decline in five of the 17 cell lines. These findings underscore the need to address Qp loss during cell line development and clone screening for recombinant GS/MSX‐CHO cell lines. Ultimately, these results ruled out cryopreservation as a potential cause for Qp decline with cell age by using contemporary non‐cryopreserved cells as controls to account for the effects of cell age.

We thank the following contributors: Cell Banking group, led by Marcia Coyne, for cryopreserving the cell lines; Beth Kao, Salina Louie, and Sun‐Ok Hwang for developing the cell lines; Analytical Operations, especially Pardis Navid and Renee Yang for performing titer and product quality assays; Amy Shen, Shahram Misaghi, and especially Jesssica Wuu for helpful input; and Andy Lin, Bob Kiss, Brad Mauger, David Shaw, John Joly, Laura Simmons, Masaru Shiratori, Mike Laird, Steve Lang, and Wendy Hsu for supporting this collaboration.

C

cryopreserved

CHO

Chinese hamster ovary

DHFR

dihdrofolate reductase

FITC

fluorescein isothiocynate

GS

glutamine synthetase

IgG

immunoglobulin G

IVCC

integrated viable cell concentration

mAb

monoclonal antibody

MFI

mean fluorescence intensity

MSX

methionine sulfoximine

MTX

methotrexate

NC

non‐cryopreserved

NS0

non‐secreting murine myeloma

pcd

pg cell−1 day−1

PBS

phosphate‐buffered saline

PDL

population doubling level

PI

propidium iodide

Qp

cell‐specific productivity

VCC

viable cell concentration

Additional Supporting Information may be found in the online version of this article.

Supporting Information

1 Schiff LJ. Review: production, characterization, and testing of banked mammalian cell substrates used to produce biological products. In Vitro Cell Dev Biol Anim. 2005;41:65–70.

• 2 Barnes LM, Bentley CM, Dickson AJ. Stability of recombinant protein production in the GS‐NS0 expression system is unaffected by cryopreservation. Biotechnol Prog. 2003;19:233–237.
• 3 Frye C, Deshpande R, Estes S, Francissen K, Joly J, Lubiniecki A, Munro T, Russell R, Wang T, Anderson K. Industry view on the relative importance of “clonality” of biopharmaceutical‐producing cell lines. Biologicals. 2016;44:117–122.
• 4 Baust JG, Gao D, Baust JM. Cryopreservation: an emerging paradigm change. Organogenesis. 2009;5:90–96.
• 5 Kim JY, Kim YG, Lee GM. CHO cells in biotechnology for production of recombinant proteins: current state and further potential. Appl Microbiol Biotechnol. 2012;93:917–930.
• 6 Nelving A, Svensson I, Lundgren M. Stability and productivity of CHO pools with respect to culture age, cryopreservation and 20 L bioreactor cultivation. In: Noll T, editor. Cells and Culture: Proceedings of the 20th ESACT Meeting, Dresden, Germany, June 17–20, 2007. Dordrecht: Springer; 2010:307–311.
• 7 Jun SC, Kim MS, Hong HJ, Lee GM. Limitations to the development of humanized antibody producing Chinese hamster ovary cells using glutamine synthetase‐mediated gene amplification. Biotechnol Prog. 2006;22:770–780.
• 8 Misaghi S, Chang J, Snedecor B. It's time to regulate: coping with product‐induced nongenetic clonal instability in CHO cell lines via regulated protein expression. Biotechnol Prog. 2014;30:1432–1440.
• 9 Bailey LA, Hatton D, Field R, Dickson AJ. Determination of Chinese hamster ovary cell line stability and recombinant antibody expression during long‐term culture. Biotechnol Bioeng. 2012;109:2093–2103.
• 10 Dorai H, Corisdeo S, Ellis D, Kinney C, Chomo M, Hawley‐Nelson P, Moore G, Betenbaugh MJ, Ganguly S. Early prediction of instability of Chinese hamster ovary cell lines expressing recombinant antibodies and antibody‐fusion proteins. Biotechnol Bioeng. 2012;109:1016–1030.
• 11 Barnes LM, Bentley CM, Dickson AJ. Stability of protein production from recombinant mammalian cells. Biotechnol Bioeng. 2003;81:631–639.
• 12 Heath C, Dilwith R, Belfort G. Methods for increasing monoclonal antibody production in suspension and entrapped cell cultures: biochemical and flow cytometric analysis as a function of medium serum content. J Biotechnol. 1990;15:71–89.
• 13 Ozturk SS, Palsson BO. Loss of antibody productivity during long‐term cultivation of a hybridoma cell line in low serum and serum‐free media. Hybridoma. 1990;9:167–175.
• 14 Park SH, Ryu DD. Cell cycle kinetics and monoclonal antibody productivity of hybridoma cells during perfusion culture. Biotechnol Bioeng. 1994;44:361–367.
• 15 Hu Z, Guo D, Yip SS, Zhan D, Misaghi S, Joly JC, Snedecor BR, Shen AY. Chinese hamster ovary K1 host cell enables stable cell line development for antibody molecules which are difficult to express in DUXB11‐derived dihydrofolate reductase deficient host cell. Biotechnol Prog. 2013;29:980–985.
• 16 Hsu WT, Aulakh RP, Traul DL, Yuk IH. Advanced microscale bioreactor system: a representative scale‐down model for bench‐top bioreactors. Cytotechnology. 2012;64:667–678.
• 17 McNeeley KM, Sun Z, Sharfstein ST. Techniques for dual staining of DNA and intracellular immunoglobulins in murine hybridoma cells: applications to cell‐cycle analysis of hyperosmotic cultures. Cytotechnology. 2005;48:15–26.
• 18 Charpa SC, Canale RP. Numerical Methods for Engineers: With Programming and Software Applications. New York: McGraw‐Hill; 1998.
• 19 Cadena‐Herrera D, Esparza‐De Lara JE, Ramírez‐Ibañez ND, López‐Morales CA, Pérez NO, Flores‐Ortiz LF, Medina‐Rivero E. Validation of three viable‐cell counting methods: manual, semi‐automated, and automated. Biotechnol Rep (Amst). 2015;7:9–16.
• 20 Yuk IH, Zhang JD, Ebeling M, Berrera M, Gomez N, Werz S, Meiringer C, Shao Z, Swanberg JC, Lee KH, Luo J, Szperalski B. Effects of copper on CHO cells: insights from gene expression analyses. Biotechnol Prog. 2014;30:429–442.
• 21 Gennaro LA, Salas‐Solano O. On‐line CE‐LIF‐MS technology for the direct characterization of N‐linked glycans from therapeutic antibodies. Anal Chem. 2008;80:3838–3845.
• 22 Wurm FM, Wurm MJ. Cloning of CHO cells, productivity and genetic stability: a discussion. Processes. 2017;5:20.
• 23 Davies SL, Lovelady CS, Grainger RK, Racher AJ, Young RJ, James DC. Functional heterogeneity and heritability in CHO cell populations. Biotechnol Bioeng. 2013;110:260–274.
• 24 Lloyd DR, Leelavatcharamas V, Emery AN, Al‐Rubeai M. The role of the cell cycle in determining gene expression and productivity in CHO cells. Cytotechnology. 1999;30:49–57. 3):
• 25 Lloyd DR, Holmes P, Jackson LP, Emery AN, Al‐Rubeai M. Relationship between cell size, cell cycle and specific recombinant protein productivity. Cytotechnology. 2000;34:59–70.
• 26 Kumar N, Gammell P, Clynes M. Proliferation control strategies to improve productivity and survival during CHO based production culture: a summary of recent methods employed and the effects of proliferation control in product secreting CHO cell lines. Cytotechnology. 2007;53:33–46.
• 27 Chusainow J, Yang YS, Yeo JH, Toh PC, Asvadi P, Wong NS, Yap MG. A study of monoclonal antibody‐producing CHO cell lines: what makes a stable high producer? Biotechnol Bioeng. 2009;102:1182–1196.
• 28 Osterlehner A, Simmeth S, Gopfert U. Promoter methylation and transgene copy numbers predict unstable protein production in recombinant Chinese hamster ovary cell lines. Biotechnol Bioeng. 2011;108:2670–2681.
• 29 Beckmann TF, Kramer O, Klausing S, Heinrich C, Thute T, Buntemeyer H, Hoffrogge R, Noll T. Effects of high passage cultivation on CHO cells: a global analysis. Appl Microbiol Biotechnol. 2012;94:659–671.
• 30 Yang Y, Mariati Chusainow J, Yap MG. DNA methylation contributes to loss in productivity of monoclonal antibody‐producing CHO cell lines. J Biotechnol. 2010;147:180–185.
• 31 Strutzenberger K, Borth N, Kunert R, Steinfellner W, Katinger H. Changes during subclone development and ageing of human antibody‐producing recombinant CHO cells. J Biotechnol. 1999;69:215–226.
• 32 Nielsen L, Borth N. Editorial: on the cusp of rational CHO cell engineering. Biotechnol J. 2015;10:929–930.
• 33 Lee JS, Grav LM, Pedersen LE, Lee GM, Kildegaard HF. Accelerated homology‐directed targeted integration of transgenes in Chinese hamster ovary cells via CRISPR/Cas9 and fluorescent enrichment. Biotechnol Bioeng. 2016;113:2518–2523.
• 34 Zhou H, Liu ZG, Sun ZW, Huang Y, Yu WY. Generation of stable cell lines by site‐specific integration of transgenes into engineered Chinese hamster ovary strains using an FLP‐FRT system. J Biotechnol. 2010;147:122–129.
• 35 Scarcelli JJ, Shang TQ, Iskra T, Allen MJ, Zhang L. Strategic deployment of CHO expression platforms to deliver Pfizer's monoclonal antibody portfolio. Biotechnol Prog. 2017;33:1463–1467.

PHOTO (COLOR): Sequence of main activities involved in generating and testing each set of eight cryopreserved (○) and eight non‐cryopreserved (×) cell lines expressing a specific mAb (i.e., mAb A, B, or C). Three sets of cell lines (first set expressing mAb A, second set expressing mAb B, and third set expressing mAb C) were generated and tested following this sequence. To generate each set of cell lines expressing a specific mAb, the CHO‐K1 parental strain was transfected with expression vectors containing the GS gene and the cDNA for the heavy and light chains of the desired mAb. The transfected cells were single‐cell cloned and screened for mAb production. Each of the top eight mAb‐producing cell lines was further processed in parallel, by cryopreserving a portion of the cells and by maintaining a portion of the remaining cells in shake flasks (constituting the continuous long‐term culture). The cryopreserved cells were thawed and maintained in shake flasks (constituting the seed train culture from cryopreserved cell lines). For each set of eight cell lines (expressing mAb A, B, or C), seed train cultures sourced from both cryopreserved and non‐cryopreserved cells were subsequently tested in three consecutive sets of production culture assays. Experiments 1, 2, and 3 (production culture assays) were conducted sequentially and each experiment was spaced approximately 1 month apart from the preceding experiment.

PHOTO (COLOR): Performance of 24 cryopreserved cell lines at three initial seed train stages—end of thaw passage (+), second passage after thaw (▵), and fourth passage after thaw (⋄)—in terms of (A) end of passage viability and (B) growth rate.

PHOTO (COLOR): Performance of 24 cryopreserved (○) and 24 non‐cryopreserved (×) cell lines in the seed train passage used to source Experiments 1, 2, and 3. These 48 seed train cultures were assessed for (A) viability and (B) growth rate. Viability data were not available for mAb A cell lines in Experiment 2. mAb C cell line #6 was not evaluated in Experiments 2 and 3 because of low product titer (<0.05 g/L for both cryopreserved and non‐cryopreserved cells) in Experiment 1.

PHOTO (COLOR): Performance of cryopreserved (○) and non‐cryopreserved (×) cell lines in terms of (A) viability and (B) IVCC at the time of harvest (day 14). Experiments 1, 2, and 3 (production culture assays) were conducted sequentially at intervals of ∼1 month (Table 1). mAb C cell line #6 was not evaluated in Experiments 2 and 3 because of low product titer (<0.05 g/L for both cryopreserved and non‐cryopreserved cells) in Experiment 1.

PHOTO (COLOR): Performance of cryopreserved (○) and non‐cryopreserved (×) cell lines in terms of (A) mAb titer and (B) Qp. Experiments 1, 2, and 3 (production culture assays) were conducted sequentially at intervals of ∼1 month (Table 1). mAb C cell line #6 was not evaluated in Experiments 2 and 3 because of low product titer (<0.05 g/L for both cryopreserved and non‐cryopreserved cells) in Experiment 1.

PHOTO (COLOR): Size distribution in terms of (A) monomer and (B) high molecular weight species for mAb produced by cryopreserved (○) and non‐cryopreserved (×) cell lines. Experiments 1, 2, and 3 (production culture assays) were conducted sequentially at intervals of ∼1 month (Table 1). Size distribution was determined by size exclusion chromatography, and was not assessed for the cases with insufficient mAb. mAb C cell line #6 was not evaluated in Experiments 2 and 3 because of low product titer (<0.05 g/L for both cryopreserved and non‐cryopreserved cells) in Experiment 1.

PHOTO (COLOR): Charge distribution in terms of (A) acidic species and (B) basic species for mAb produced by cryopreserved (○) and non‐cryopreserved (×) cell lines. Experiments 1, 2, and 3 (production culture assays) were conducted sequentially at intervals of ∼1 month (Table 1). Charge distribution was determined by imaged capillary isoelectric focusing, and was not assessed for the cases with insufficient mAb. mAb C cell line #6 was not evaluated in Experiments 2 and 3 because of low product titer (<0.05 g/L for both cryopreserved and non‐cryopreserved cells) in Experiment 1.

PHOTO (COLOR): Glycosylation distribution in terms of (A) G0 species and (B) Man5 species for mAb produced by cryopreserved (○) and non‐cryopreserved (×) cell lines. Experiments 1, 2, and 3 (production culture assays) were conducted sequentially at intervals of ∼1 month (Table 1). Glycosylation distribution was determined by capillary electrophoresis, and was not assessed for the cases with insufficient mAb. mAb C cell line #6 was not evaluated in Experiments 2 and 3 because of low product titer (<0.05 g/L for both cryopreserved and non‐cryopreserved cells) in Experiment 1.

PHOTO (COLOR): Flow cytometric analysis of intracellular IgG expression from seed train cultures. mAb A cell line #2 was sourced from (A) cryopreserved and (B) non‐cryopreserved cells. mAb B cell line #2 was sourced from (C) cryopreserved and (D) non‐cryopreserved cells. The cells were fixed, permeabilized, and incubated with FITC conjugated anti‐human IgG. Flow cytometry was performed on cells used for initiating (i) Experiment 1, (ii) Experiment 2, and (iii) Experiment 3. The histograms represent normalized cell counts vs. fluorescence intensity (on a log scale) for negative control (non‐transfected cells in red) and mAb cell line (in blue).