High-Throughput, Automated Protein A Purification ... - ACS Publications

Aug 3, 2016 - (1) Rathore, A. S.; Winkle, H. Nat. ... B.; DeFelippis, M. R.; Rudd, P. M. Anal. ... Chance, M. R.; Sali, A.; Acton, T. B.; Xiao, R.; Zh...
0 downloads 0 Views 4MB Size
This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Article pubs.acs.org/ac

High-Throughput, Automated Protein A Purification Platform with Multiattribute LC−MS Analysis for Advanced Cell Culture Process Monitoring Jia Dong,*,† Nicole Migliore,‡ Steven J. Mehrman,§ John Cunningham,∥ Michael J. Lewis,† and Ping Hu*,† †

Large Molecule Analytical Development, Pharmaceutical Development & Manufacturing Science, Janssen Research & Development, Malvern, Pennsylvania 19355, United States ‡ Active Pharmaceutical Ingredient Large Molecule, Pharmaceutical Development & Manufacturing Science, Janssen Research & Development, Malvern, Pennsylvania 19355, United States § Strategic Operations, Pharmaceutical Development & Manufacturing Science, Janssen Research & Development, Spring House, Pennsylvania 19477, United States ∥ Material Science, Pharmaceutical Development & Manufacturing Science, Janssen Research & Development, Malvern, Pennsylvania 19355, United States S Supporting Information *

ABSTRACT: The levels of many product related variants observed during the production of monoclonal antibodies are dependent on control of the manufacturing process, especially the cell culture process. However, it is difficult to characterize samples pulled from the bioreactor due to the low levels of product during the early stages of the process and the high levels of interfering reagents. Furthermore, analytical results are often not available for several days, which slows the process development cycle and prevents “real time” adjustments to the manufacturing process. To reduce the delay and enhance our ability to achieve quality targets, we have developed a low-volume, high-throughput, and high-content analytical platform for at-line product quality analysis. This workflow includes an automated, 96-well plate protein A purification step to isolate antibody product from the cell culture fermentation broth, followed by rapid, multiattribute LC−MS analysis. We have demonstrated quantitative correlations between particular process parameters with the levels of glycosylated and glycated species in a series of small scale experiments, but the platform could be used to monitor other attributes and applied across the biopharmaceutical industry.

P

advanced control equipment, and mathematical and statistical modeling.8−12 Real-time monitoring of several key variables has been implemented for control of the cell culture process, including pH, temperature, gas-phase oxygen, carbon dioxide, viable cell density, osmolality, metabolites, and titer.13−16 However, significant challenges still exist because cell culture systems are very complex.17,18 The host cell line, clone, media composition, feeding strategy, and fermentation conditions

roduct and process understanding are key elements of the Quality by Design (QbD) initiative outlined by the FDA.1,2 QbD implies that quality has been built into the manufacturing process at the design stage. In addition, improved online monitoring and control methods should be implemented to maintain high product quality during the manufacturing process.3 To help realize these goals, process analytical technology (PAT) is being used to identify critical process parameters (CPP) that are needed to control the levels of critical quality attributes (CQA).4−7 A number of PAT tools have been developed in recent years to improve the control of product quality, including new data acquisition methodologies, © XXXX American Chemical Society

Received: May 19, 2016 Accepted: August 3, 2016

A

DOI: 10.1021/acs.analchem.6b01956 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Table 1. Summary of 2(5‑1) Factorial Cell Culture Design of Experiment (DOE)a

impact the levels of many protein quality attributes such as glycosylation, glycation, oxidation, and aggregation,19 and these product attributes are difficult to monitor in a timely manner. Protein glycosylation is especially sensitive to cell culture conditions and changes in glycosylation can have significant impact on antibody-dependent cell-mediated cytotoxicity (ADCC) activity, complement-dependent cytotoxicity (CDC) activity, clearance, and immunogenicity.20 Thus, controlling protein glycosylation is particularly important for ensuring product quality and consistency. Currently, product characterization is typically performed postharvest. This type of single-point measurement can only reflect the cumulative impact of many process variables on final product quality and cannot assist in real-time process feedback control. Therefore, developing high-throughput in-line or atline analytical tools is critical to identify correlations between process parameters and product quality attributes to help advance process understanding, improve product quality, and increase production efficiency. To date, only a few approaches have been developed targeting at-line or in-line control of product quality for protein production.21 For example, a method was developed to analyze antibody charge variants using 2D-UPLC with automatic sample collection from the bioreactor.22 Antibody aggregation kinetics were measured using light scattering of up to eight samples with low detection limit, which is promising for online measuring of aggregation under different conditions or liquid formulations.23 An automated robotic system was developed to analyze antibody glycosylation directly from supernatants by glycan release, fluorescent labeling, purification, and subsequent HILIC analysis.24 Subunit LC−MS has also been used to analyze glycosylation on cell culture supernatant samples.25 Although fast and easy, high noise levels interfered with MS detection and subsequent data analysis, most likely due to interference from amino acids, lipids, and other media components. Thus, a method to purify proteins from low volumes of fermentation broth is needed in order to develop highly sensitive analytical assays for at-line control of the cell culture process. In this study, a high-throughput, automated platform was developed for monitoring multiple product attributes from low volumes of fermentation broth. This workflow included protein A purification of clarified cell culture harvest in a 96-well plate format using the Tecan Freedom EVO workstation, followed by rapid LC−MS analysis. The workflow was used to monitor aglycosylation, glycosylation, and glycation but could be extended to other product attributes as well. The applicability of this workflow was demonstrated by a series of experiments at the edges of standard control limits to purposely induce variation in the product quality attributes throughout the fermentation process.

run no. 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 3.1 a

pH

dissolved oxygen

− + − + − − + + + − + − + + − −

− + + − + − − + − − + − − + + +

glucose target + + − + − + − − − − + − + − + + control

temperature

seeding density

− − + − − + + + − + + − + − − +

+ + − − + − − + + + − − + − − +

The +/− indicates set points above or below the points in control.

solution preparations. Seeding density targets were achieved by variable inoculums at the initiation of the experiment. The pH, dissolved oxygen (DO), temperature, and feeding strategy were identical for all experiments through culture Day 3. The described increases or decreases for pH and DO were implemented on Day 3. The temperature control was implemented on Day 5. Finally, a change in the feeding strategy was implemented to achieve the prescribed glucose target on Day 8. The variables were monitored in real time and maintained at the predetermined points. Daily sampling of bioreactors was conducted for routine at-line testing and additional samples were retained for LC−MS characterization. Protein A Purification and Sample Cleanup. Protein A purification was performed on Tecan Freedom EVO workstation with pre-programmed scripts followed by sample cleanup. Protein A HP multitrap plates (GE Healthcare, U.K.) were first equilibrated with 250 μL of equilibration buffer (50 mM sodium phosphate, 150 mM sodium chloride, pH 7.4). The samples (200−600 μL) were then loaded into the plate and mixed twice during a time period of 6 min, washed with 200 μL of Buffer 1 (50 mM sodium phosphate, 150 mM sodium chloride, pH 7.4), 200 μL of Buffer 2 (50 mM sodium phosphate, 1 M sodium chloride, pH 7.4), 200 μL of Buffer 3 (100 mM sodium acetate, pH 5.5), and finally eluted twice with 125 μL of 100 mM sodium acetate, pH 3.8. Protein recovery was monitored by UV absorbance at 280 nm, and samples were buffer exchanged into 100 mM Tris pH 7.5 using PD Multitrap G-25 plates (GE Healthcare, U.K.) for subsequent chemical reaction and enzymatic digestion. Ä KTA Protein Purification. Ä KTA Avant 25 system (GE Healthcare, U.K.) was used for end of batch purification of cell culture harvest in selected bioreactor runs. Prepacked HiScreen MabSURE column was used (GE Healthcare, U.K.) with 30 mg/mL load targeted for all runs. The same wash buffers and elution buffer were used for Ä KTA and Tecan-assisted protein A purification. Reduced and Subunit Mass Analysis. For reduced mass analysis, 100 μL of Protein A purified sample was reduced with 50 mM DTT for 15 min at 37 °C and quenched with 1 M HCl.



EXPERIMENTAL SECTION Cell Line, Media, and Expansion. Recombinant monoclonal antibodies (mAbs) were expressed in a Chinese Hamster Ovary (CHO) cell line using proprietary basal and feed media in 2 L DasGip bioreactors. A total of 17 conditions were evaluated using a 2(5−1) factorial design in three batches, with eight conditions in DOE 1, eight conditions in DOE 2, and one control (Table 1). The set points in DOEs were designed around the upper and lower edges from an established process for this cell line (i.e., control) in order to induce changes in the productivity and product quality attributes. The three batches were conducted with separate seed expansion and media B

DOI: 10.1021/acs.analchem.6b01956 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Automated high-throughput workflow for cell culture process monitoring.

For subunit mass analysis, 100 μL of Protein A purified sample was incubated with 0.5 units of IdeS and EndoS (Genovis AB, Sweden)/μg protein for 1 h at 37 °C, reduced with 50 mM DTT for 15 min at 37 °C, and quenched with 1 M HCl. All samples were stored at 4 °C for less than 18 h before analysis. Liquid Chromatography−Mass Spectrometry Analysis. A Waters Xevo G2-XS Q-TOF mass spectrometer coupled with a Waters Acquity I-class ultraperformance liquid chromatography system (Milford, MA) was used for all LC− MS analysis. Separation was performed on a Waters BEH300 C4 2.1 mm × 50 mm column with a gradient of acetonitrile in 0.1% formic acid at 60 °C. The column was equilibrated in 10% B, and the gradient was increased to 25% B over 2 min, to 40% B over 7 min, and then to 90% B over 0.5 min. The capillary voltage was kept at 2.5 kV, and the cone voltage was set at 40 V. The source temperature and desolvation temperature were set at 100 and 350 °C, respectively.



RESULTS AND DISCUSSION High-Throughput Characterization Workflow. A highthroughput protein A purification step using Tecan robotics in a 96-well plate format followed by buffer exchange was employed prior to LC−MS analysis of cell culture harvest samples to improve data quality (Figure 1). Generally, 96 samples were processed in a single 60 min run and protein recovery was ∼80% with mass loads ranging from 0.3 to 0.7 mg. Protein A purified samples were analyzed using two LC−MS assays to provide maximum information on product quality attributes with the shortest sample preparation and analysis time. Reduced mass analysis was used for accurate identification and quantitation of HC glycoforms following separation of the HC and LC with a 10 min RP-UPLC method. As shown in Figure 2A, this method detected the major G0F, G1F, and G2F glycoforms as well as low-abundance high-mannose species. This method was simple, fast, and generally applicable to different IgG subclasses with only minor modifications. Subunit mass analysis was used for accurate identification and quantitation of aglycosylated, afucosylated, and glycated scFc species. This method is gaining increasing interest in antibody characterization due to the high specificity, high yield, and robustness of the digestion procedure.26−28 Digestion of samples with EndoS and IdeS followed by reduction with DTT was used to produce three antibody domains (LC, Fd′, and scFc), which can be separated for LC−MS analysis.29,30 As shown in Figure 2B, aglycosylated scFc, afucosylated scFc, and scFc could be quantified from the deconvoluted mass spectrum for the scFc. In this case because EndoS cleaved the N-glycans

Figure 2. LC−MS analysis of purified cell culture samples: TIC for reduced mass analysis and HC deconvolution (panel A) and TIC for subunit mass analysis and scFc deconvolution (panel B). CTL: Cterminal lysine.

after the first GlcNAc residue, the afucosylated glycans could be quantified. Method Sensitivity and Robustness. To evaluate method sensitivity, the levels of six glycoforms (Man5, G0FGlcNAc, G0, G0F, G1F, and G2F) as well as aglycosylated, afucosylated, and glycated species were quantitated in the Day 0 samples. Results from runs 1.1−1.8 and 2.1−2.8 were grouped together, respectively, for statistical analysis because C

DOI: 10.1021/acs.analchem.6b01956 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry samples taken early in the cell culture process were expected to have similar distributions of glycoforms and other PTMs. As shown in Table S1, major and minor species were detected at levels ranging from 0.1% for Man5 to 58.0% for G0F. RSD values for most glycoforms were