- Email: [email protected]
Quantification of residual AEBSF-related impurities by reversed-phase liquid chromatography
Journal of Chromatography B 1116 (2019) 19–23
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/jchromb
Quantification of residual AEBSF-related impurities by reversed-phase liquid chromatography☆
T
Cindy X. Cai, Nicole A. Schneck, Doug Harris, Daniel Blackstock, Vera B. Ivleva, ⁎ Kuang-Chuan Cheng, Adam Charlton, Frank J. Arnold, Jonathan W. Cooper, Q. Paula Lei Vaccine Production Program, Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Gaithersburg, MD, USA
ARTICLE INFO
ABSTRACT
Keywords: RPLC-UV AEBSF AEBS-OH Protease inhibitor Antibody Clipping Impurity Residual
During research of a broadly neutralizing antibody (bNAb) for HIV-1 infection, site-specific clipping was observed during cell culture incubation. Protease inhibitor, 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF), was supplemented to the cell culture feeding to mitigate clipping as one of the control strategies. It led to the need and development of a new assay to monitor the free AEBSF-related impurities during the purification process. In this work, a reversed-phase liquid chromatography (RPLC-UV) method was developed to measure the total concentration of AEBSF and its major degradant product, 4-(aminoethyl) benzenesulfonic acid (AEBS-OH). This quantitative approach involved hydrolysis pre-treatment to drive all AEBSF to AEBS-OH, a filtration step to remove large molecules, followed by RPLC-UV analysis. The method was qualified and shown to be capable of measuring AEBS-OH down to 0.5 μM with good accuracy and precision, which was then applied for process clearance studies. The results demonstrated that a Protein A purification step in conjunction with a mock ultrafiltration/diafiltration (UF/DF) step could remove AEBSF-related impurities below the detection level. Overall, this study is the first to provide a unique approach for monitoring the clearance of free AEBSF and its related degradant, AEBS-OH, in support of the bNAb research.
1. Introduction Monoclonal antibodies (mAbs) have been long considered for various therapeutic applications due to their capability to recognize and to provide protection against foreign invading antigens [1]. Currently, they have been applied for the treatment of a variety of diseases [2]. To support the growing needs for therapeutic mAbs, the biopharmaceutical industry has been constantly improving processes to achieve products with high yields and good quality [3]. One of the key challenges for biopharmaceutical research and development is the protein product susceptibility to proteolytic clipping during the cell culture process, leading to a negative effect on biological activity [4,5]. Various approaches have been reported to solve this issue, including eliminating the enzyme reaction site through mutation of the known proteaseclipped amino acid site(s), or minimizing the protease reactions through reduction of protease concentration or their activities [6,7].
For example, adding protease inhibitor(s) to the cell culture process can impede protease actions and result in the reduction of proteolytic clipping [8,9]. However, the protease inhibitor(s) can degrade and accumulate if added continuously during the cell culture period. Therefore, it is critical to remove the protease inhibitor(s) during the purification process. In addition, a reliable quantitative method to monitor the residual protease inhibitor-related compounds must be applied to ensure product purity and safety. In a previous study, site-specific clipping within the CDR-H3 region of a broadly neutralizing antibody (bNAb) for HIV-1 infection was discovered when a fed-batch cell culture process was employed [10]. To reduce the clipping percentage, one of the approaches during the research stage involved the application of a protease inhibitor, 4-(2aminoethyl) benzenesulfonyl fluoride (AEBSF), to the cell culture bioreactor. Besides binding to proteases to prevent proteolysis, AEBSF could hydrolyze into 4-(aminoethyl) benzenesulfonic acid (AEBS-OH)
Abbreviations: AEBSF, 4-(2-aminoethyl) benzenesulfonyl fluoride; AEBS-OH, 4-(aminoethyl) benzenesulfonic acid; TFA, trifluoroacetic acid; mAb, monoclonal antibody; bNAb, broadly neutralizing antibody; RPLC-UV, reversed-phase liquid chromatography; UV, ultraviolet/visible; CV, column volume; RSD, relative standard deviation; PBS, phosphate-buffered saline; MWCO, molecular weight cut-off:; QC, quality control; LLOQ, lower limit of quantification; UF/DF, filtration ultrafiltration and diafiltration ☆ Declaration of interest: None. ⁎ Corresponding author at: 9 West Watkins Mill Rd, Gaithersburg, MD 20878, USA. E-mail address: [email protected] (Q.P. Lei). https://doi.org/10.1016/j.jchromb.2019.03.022 Received 19 December 2018; Received in revised form 19 February 2019; Accepted 19 March 2019 Available online 20 March 2019 1570-0232/ Published by Elsevier B.V.
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
2.3. Amicon™ filtration purification To represent potential clearance of AEBSF through tangential flow filtration ultrafiltration and diafiltration (UF/DF), samples were purified using Amicon centrifuge filters with a 30 kDa MWCO. Protein A purified bNAb was spiked with 0.1 mg/mL AEBS-OH and then centrifuged using Amicon 15 mL-ultra filters for 20 min at 4000g. After each centrifugation, 7 mL of fresh diafiltration buffer was added to the Amicon filter to represent additional diafiltration volumes. This process was repeated 5 times before collecting the final retentate product. 2.4. Sample preparation for RPLC-UV analysis Scheme 1. Hydrolysis reaction of AEBSF.
All residual AEBSF were fully hydrolyzed to form one “product” peak for RPLC-UV quantitative purposes. Based on previously study, AEBS-OH stock solution was prepared by incubating 500 μM AEBSF in 100 mM Tris buffer (pH 12.0) at 37 °C for 2 h [11]. The stock standard solution was further diluted accordingly using 100 mM Tris buffer to obtain the following six concentrations: 0.5, 1, 5, 10, 15 and 20 μM. These AEBS-OH standards were subsequently used as calibrants to generate a linear standard curve. A system suitability/quality control (QC) sample was prepared by diluting AEBSF in water to a concentration of 20 μM. QC and test samples were then mixed with 200 mM Tris buffer (pH adjusted to 12.0 using sodium hydroxide) at 1:1 (v:v) ratio followed by incubation at 37 °C for 2 h to ensure complete conversion of ABESF to AEBS-OH. The final QC concentration for the analyte was 10 μM, which matched the middle point of the calibration curve. Before analysis, 200 μL of each ABESF-OH calibrant, QC and test sample were filtered through Amicon filters (10 kDa MWCO) for 30 min either at 3220 g using an Eppendorf centrifuge (model 5810r, Hauppauge, NY) or at 16162 g using a Beckman centrifuge system (model Coulter Microfuge 16, Indianapolis, Indiana), which were the highest speed settings of the centrifuges. It ensured the minimum retention volume in the filter to reduce variability caused by volume differences. This process also separated AEBS-OH (202 Da) from the large molecules prior to RPLC-UV analysis. Duplicates of each standard and sample were prepared during each run. The QC sample was injected before and after all samples as a bracketing control.
at a certain rate during the cell culture process (Scheme 1) [11]. Subsequent purification steps were then applied to remove AEBSF and its degradation product. Herein, we discuss the development of a reversedphase liquid chromatography (RPLC) method with ultraviolet/visible (UV) detection to quantify the process-related impurities, AEBSF and its degradation product, AEBS-OH. In this method, samples were first hydrolyzed, during which AEBSF was completely driven to AEBS-OH (AEBSF hydrolyzed form). The hydrolyzed sample was then filtered to separate AEBS-OH (202 Da) from larger molecules (e.g., bNAb) to minimize matrix interferences and to achieve quantification of the residual impurity. This newly developed RPLC-UV approach demonstrated good sensitivity and accuracy, which was subsequently applied to support process clearance efforts by monitoring AEBSF-related impurities at various stages of purification. 2. Materials and methods 2.1. Chemicals Chemicals were analytical grade unless specified. Acetic acid, HPLC grade water and HPLC grade acetonitrile were purchased from J. T. Baker (Center Valley, PA). Tris buffer and HPLC grade trifluoroacetic acid (TFA) were purchased from Thermo Fisher Scientific (Tempe, AZ). AEBSF hydrogen chloride was purchased from G-Bioscience (St. Louis, MO). Phosphate-buffered saline (PBS) was purchased from Hyclone (Logan, UT). Sodium hydroxide was purchased from VWR (Radnor, PA), and ethanol was purchased from Warner-Graham Company (Cockeysville, MD). The Amicon Ultra centrifugal filters (0.5 mL or 15 mL capacities) with molecular weight cut-offs (MWCOs) of 10 kDa or 30 kDa were purchased from Millipore Sigma (Burlington, MA).
2.5. RPLC-UV method set up The Acquity H-class Bio UPLC system (Waters, MA), consisting of a quaternary solvent manager, sample manager and UV detector, was operated using Empower v4.0. Chromatographic separation was performed on an AdvanceBio Peptide Mapping C18 column (Agilent, 2.7 μm, 2.1 × 250 mm), at ambient temperature and injection volume of 50 μL. The detection wavelength was set at 220 nm, which was optimized using the maximum UV-absorbance after screening in the range of 190–600 nm with UV spectrometry. Mobile phase A consisted of 0.1% (v/v) TFA in water, and mobile phase B consisted of 0.085% TFA in acetonitrile. Mobile phases were delivered at a flow rate of 0.15 mL/ min using an isocratic gradient of 100% mobile phase A for 8 min. After elution of AEBS-OH, the gradient was quickly ramped up to 95% mobile phase B and held for 10 min to wash the column, followed by column re-equilibration (100% mobile phase A) for 5 min prior to the next injection. Each run was 23 min.
2.2. Protein A purification To demonstrate clearance of AEBSF after Protein A purification, AEBSF was spiked into Protein A purified bNAb to reach the concentration around 1 mg/mL (~4000 μM) in the matrix of cell culture protein A flow-through or Tris buffer. These mock samples, as well as the cell culture harvest supplemented with AEBSF during cell culture were purified over 600 μL Protein A RoboColumns (Toyopearl AFrProtein A HC-650F) using the Tecan liquid handling platform. The system was equilibrated with 5 column volumes (CVs) of PBS before sample loading, followed by two sequential washing steps with 5 CVs solution in each. Protein was eluted at low pH and neutralized using 1 M Tris (i.e. Protein A elution). The column was cleaned and regenerated for next use. All steps were performed using a flow rate of 1.9 μL/s.
2.6. RPLC-UV method qualification The AEBS-OH calibration curve was constructed by plotting detected UV peak area versus known concentration of AEBS-OH. The analyte concentration in the test samples was determined by back-calculating sample peak areas against the calibration curve, accounting for the dilution factor. Quality characteristics of the method, including specificity, linearity, sensitivity, accuracy, precision and sample 20
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
stability were evaluated. Method specificity was assessed by evaluating matrix interference potentially caused by the Tris buffer, a Protein A purified bNAb sample, or mock cell culture harvest (cell culture flow-through spiked with bNAb but without AEBSF or AEBS-OH). Method linearity was assessed by linear regression (R2) of AEBS-OH standard curves. The accuracy of the assay was evaluated by determining the percent recovery of AEBSOH concentration detected versus the spiked at 2, 10, and 18 μM into Protein A purified bNAb samples. The method intermediate precision range was tested using two bNAb samples, pre-spiked with AEBSF (~4000 mM) in mock cell culture harvest followed by Protein A purification. Samples were analyzed by variable factors, including two analysts, two columns, and two centrifuges in three different days. Lastly, for stability analysis of hydrolyzed samples, the concentration of AEBS-OH in the QC and two bNAb Protein A purified samples was obtained at timepoints 0 h and 44 h against the standard curve created at time 0 h. The detected concentration difference was evaluated to determine sample stability after hydrolysis.
Fig. 2. RPLC-UV chromatograms (y-axis normalized) for AEBSF and AEBS-OH spiked (a) in Tris buffer (b) followed by Amicon filtration, and (c) in mock cell culture harvest followed by Protein A purification and Amicon filtration.
AEBS-OH (Fig. 2). However, the AEBSF peak co-eluted with unknown matrix interference peaks from the Protein A purified sample (Fig. 2) even after a few attempts of method optimization. It imposed a challenge on direct quantification of AEBSF, the process related impurity; therefore, an alternative approach was developed. Because AEBS-OH was well-separated from other interferences and is a naturally degraded product of AEBSF during process, a strategy to drive all AEBSF into its hydrolyzed form, AEBS-OH, was employed for the quantification study. This approach allowed measurement of a single analyte for straightforward RPLC-UV analysis and quantification. More than 99% AEBSF converted to AEBS-OH after a 2-h incubation at 37 °C within the pH range of 9.0–12.0 (Fig. S2 in supplementary). Since the pH of the Protein A purified samples typically ranged from 6.0 to 7.6 after neutralization, samples were mixed with Tris buffer at a higher pH 12.0 to ensure AEBSF was fully hydrolyzed at pH 9.0 and above. The hydrolyzed samples were then filtered to reduce large proteins to extend the column lifetime. Using this sample pretreatment scheme, detected AEBS-OH using the newly developed RPLC-UV method accounted for the total amount of AEBSF and its degradation product, AEBS-OH.
3. Results and discussion 3.1. RPLC-UV method development To reduce the site-specific proteolytic clipping of the bNAb during cell culture incubation, AEBSF was supplemented into the fed-batch cell culture media daily at the concentration of 500 μM from day 7–14 (8 days). Besides binding to proteases, AEBSF also hydrolyzed into AEBS-OH as described in a previous study [11]. A proof-of-concept experiment in the supplementary demonstrated that most of AEBSF converted to AEBS-OH, but 9.1% AEBSF stayed in its original form after a 24-h hydrolysis in cell culture-like conditions at pH 7.0 and 37 °C in Tris buffer (Fig. S1 in supplementary). This indicated that AEBSF and AEBS-OH potentially coexisted in the cell culture harvest after the last AEBSF feed (24 h before harvest) to the cell culture media. Therefore, measurement of both in-process related impurities, AEBSF and AEBSOH, was required. However, no previous quantification method has been reported for analysis of both compounds; thereby, we address the need to develop a method to quantify the total amount of AEBSF and AEBS-OH. Both compounds contain benzene ring with different functional groups, therefore, a C18 stationary phase can interact with the benzene ring, retain and separate them according to their different hydrophobicity [12]. At the initial method development stage, various C18 columns with different stationary phases were screened to retain the highly hydrophilic molecule, AEBS-OH. The AdvanceBio column was thereafter chosen since it successfully retained AEBS-OH at around 6 min and AEBSF at around 18 min (Fig. 1), which demonstrated the good separation of AEBSF and AEBS-OH. Extraneous peaks were observed after Amicon filtration but did not interfere with either AEBSF or
3.2. Assay quality assessment The specificity of the method was assessed by testing various diluents and buffers (i.e., water, Tris buffer and PBS) applied during sample handling, as well as the test sample matrices (i.e., Protein A purified bNAb sample and 100× diluted mock cell culture harvest). No matrix-related peaks were observed around the AEBS-OH elution time (Fig. S3 in supplementary). Additionally, salt in the buffer was eluted in the first 3 mins and other compounds with higher hydrophobicity eluted after 8 min (when the percentage of organic mobile phase increased) even in the most complicated matrix, such as mock cell culture harvest. Thus, no background interference was observed within the retention time of 5.5–7.5 min, providing the high specificity of the method. Linearity of the method was assessed in the range of 0.5–20 μM of AEBS-OH in Tris buffer. A representative standard curve is shown in Fig. 3. Five individually prepared AEBS-OH calibration curves all generated linear correlations of R2 > 0.99, demonstrating a good linear response. Sensitivity of the method was established based on the lower limit of quantification (LLOQ) of AEBSF standard curve at 0.5 μM (~100 ng/ mL), of which signal-to-noise (S/N) was above 10 among the 20 replicate injections. Accuracy was inferred by AEBS-OH recovery during spiking experiments. AEBSF spiked into the Protein A purified bNAb samples could potentially bind to the bNAb [13], but monitoring the modified bNAb-AEBSF was beyond the testing scope of this method. Therefore,
Fig. 1. RPLC-UV chromatograms (y-axis normalized) of (a) AEBSF and (b) AEBS-OH standards in Tris buffer (final concentration: 20 μM). 21
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
Table 3 Sample stability analysis of hydrolyzed AEBS-OH QC and Protein A purified bNAb samples at storage conditions up to 44 h. Samples
Detected AEBS-OH (μM) 0h
QC Protein A purified samples
Difference %
44 h 9.8 6.6 41.3
10.4 6.9 43.7
6.7 5.0 5.8
was able to accurately quantify the total amount of free AEBSF and AEBS-OH in solution down to 0.5 μM, with the accuracy within the acceptance precision range. 3.3. Application of the RPLC-UV method for clearance study using protein A purification
Fig. 3. A representative AEBS-OH standard curve in the detection range of 0.5–20 μM.
Protein A capture as the initial downstream purification step removed the majority of in-process related impurities. A clearance study was applied to evaluate the mass balance and clearance power of free AEBSF-related impurities, which was essential for understanding the Protein A purification removal capability and limitations [16]. Two samples were prepared (details in supplementary) for (1) a proof-ofconcept study, a purified bNAb sample spiked with AEBSF (~4000 μM), which represented the maximum amount of AEBSF added to the bioreactor during the cell culture process, and (2) the real case study, the cell culture harvest with AEBSF supplement strategy. Both samples were purified using a Protein A column. The Protein A flow-through, wash, and purified bNAb elution were all collected. The concentration of AEBSF-related impurities was detected in each fraction, as well as in the loading materials. The mass balance for the purification step was defined as input(load) = output(elution) + removal(flow-through and wash) [17], which helped track the AEBSF-related impurities at various stages in the purification system. The input was defined as the total amount of AEBSF-related compounds loaded onto the Protein A column, while the output (i.e. Protein A elution) was defined as the total amount of AEBSF-related impurities measured in the Protein A purified bNAb sample. The total mass recovery of AEBSF and AEBS-OH was calculated by (output(elution) + removal (flow-through and wash)) / input(load) × 100%. Recovery was 76.7% and 101.6% in both case studies (Table 4). The results also showed that AEBSF and AEBS-OH was removed mainly during the column flow-through step and slightly during the following wash step (Table S1 and Table S2 in supplementary), but residual amount of free AEBSF and AEBS-OH still remained in the Protein A purified bNAb sample. The normalized AEBSF and AEBS-OH amount against the bNAb amount was obtained at 1.00 and 0.123 μg per mg bNAb in two case studies (Table 4), respectively. To this end, a clearance power of AEBSF and AEBS-OH was determined by input(load) / output(elution) × 100%, which was 179-fold and 33-fold in two different case scenarios (Table 4), a proof-of concept experiment and a real cell culture harvest purification process, respectively.
Table 1 Method accuracy analysis by percent recovery of AEBS-OH spiked into Protein A purified bNAb samples. AEBS-OH (μM) Spiked
Detected
2.0 10.0 18.0
2.0 9.3 17.6
Average recovery%
RSD% (n = 4)
93.8 98.3 95.1
13.3 6.7 3.9
only AEBS-OH was spiked into the Protein A purified bNAb samples at different concentration for accuracy analysis. The recovery of AEBS-OH was between 93 and 100% with precision RSD < 10% (Table 1). Thus, good method accuracy was achieved. Precision was evaluated by analyzing the following samples. About 4000 μM of AEBSF were spiked into mock cell culture harvest in two experiments, which mimicked the maximum amount of AEBSF addition to the cell culture bioreactor. Both spiked samples were purified using a Protein A column and then tested using the newly developed RPLC-UV method. The method intermediate precision was assessed on three days (duplicate sample preparations for each sample on each run) by two different analysts, using two different columns and two different centrifuge systems. As shown in Table 2, the relative standard deviation (% RSD) was < 20% with a total of 6 runs performed. The precision was within the acceptable range for residual impurity analysis [14,15]. The stability of AEBS-OH in Tris buffer and Protein A purified samples after hydrolysis was evaluated at time-points 0 h and 44 h under a storage condition of 2–8 °C. UV peak areas of the QC sample and two Protein A purified samples were monitored at the two timepoints (0 h and 44 h), and then interpolated against the standard curve prepared at the time-point 0 h. As demonstrated in Table 3, the AEBSOH concentrations measured at time-point 0 h and 44 h had a percent difference of < 10%, which was within the acceptance range. Therefore, hydrolyzed samples stored at 2–8 °C were deemed stable and acceptable for analysis for at least 44 h, which enabled > 100 injections to be batched into a single run with high confidence. Therefore, the pre-hydrolysis step followed by RPLC-UV analysis
3.4. Application of the RPLC-UV method for clearance study using Amicon filtration UF/DF can separate small molecules from large proteins; therefore, this additional purification step has the potential to further remove AEBSF (202 Da) and AEBS-OH (204 Da) in the bNAb (about 150 kDa) product. A proof-of-concept experiment was designed at the small scale using Amicon filtration to mimic the UF/DF step. AEBS-OH was spiked into a Protein A purified bNAb sample to reach the concentration of 8.47 μg AEBS-OH per mg bNAb, which was > 8 times higher than the concentration of free AEBSF-related impurities after Protein A purification clearance study described above (1.00 and 0.123 μg/mg in
Table 2 Intermediate precision assessment of 6 runs of AEBS-OH in two separated Protein A purified bNAb samples. Protein A purified samples 1 2
AEBSF detected in protein A purified bNAb (μM)
RSD% (n = 6)
6.0 40.4
17.3 5.6
22
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
Table 4 Clearance study results of AEBSF-related impurities for Protein A purification. Case study for AEBSF and AEBS-OH
Recovery (%)
Normalized concentration (μg/mg bNAb) in Load
(1) Proof-of-concept case study (2) Real case study
1.79 × 10 4.10
Load After filtration
1.00 0.123
179-fold 33-fold
Appendix A. Supplementary data
Table 5 Clearance study results for Amicon filtration. Samples
Protein A purified bNAb 2
76.6 101.6
Clearance power
AEBS-OH (μg/mg bNAb)
Clearance power
8.47 < 0.015
> 564-fold
Supplementary data to this article can be found online at https:// doi.org/10.1016/j.jchromb.2019.03.022. References [1] A.B. Mahmuda, F.; Al-Zihiry, K.J; Abdulhaleem, N; Majid, R.A.; Hamat, R.A.; Abdullah, W.O.; Unyah Z. , Monoclonal antibodies: a review of therapeutic applications and future prospects, Trop, J. Pharm. Res. 16 (2017) 713–722. [2] X. Geng, X. Kong, H. Hu, J. Chen, F. Yang, H. Liang, X. Chen, Y. Hu, Research and development of therapeutic mAbs: an analysis based on pipeline projects, Hum Vaccin Immunother 11 (2015) 2769–2776. [3] P. Gronemeyer, R. Ditz, J. Strube, Trends in upstream and downstream process development for antibody manufacturing, Bioengineering (Basel) 1 (2014) 188–212. [4] F. Grosse-Holz, L. Madeira, M.A. Zahid, M. Songer, J. Kourelis, M. Fesenko, S. Ninck, F. Kaschani, M. Kaiser, R.A.L. van der Hoorn, Three unrelated protease inhibitors enhance accumulation of pharmaceutical recombinant proteins in Nicotiana benthamiana, Plant Biotechnol. J. 16 (2018) 1797–1810. [5] F. Robert, H. Bierau, M. Rossi, D. Agugiaro, T. Soranzo, H. Broly, C. MitchellLogean, Degradation of an Fc-fusion recombinant protein by host cell proteases: identification of a CHO cathepsin D protease, Biotechnol. Bioeng. 104 (2009) 1132–1141. [6] H. Laux, S. Romand, S. Nuciforo, C.J. Farady, J. Tapparel, S. Buechmann-Moeller, B. Sommer, E.J. Oakeley, U. Bodendorf, Degradation of recombinant proteins by Chinese hamster ovary host cell proteases is prevented by matriptase-1 knockout, Biotechnol. Bioeng. 115 (2018) 2530–2540. [7] B.J. Ryan, G.T. Henehan, Avoiding proteolysis during protein purification, Methods Mol. Biol. 1485 (2017) 53–69. [8] S. Chakrabarti, C.J. Barrow, R.K. Kanwar, V. Ramana, J.R. Kanwar, Studies to prevent degradation of recombinant fc-fusion protein expressed in mammalian cell line and protein characterization, Int. J. Mol. Sci. 17 (2016). [9] C. Goulet, M. Benchabane, R. Anguenot, F. Brunelle, M. Khalf, D. Michaud, A companion protease inhibitor for the protection of cytosol-targeted recombinant proteins in plants, Plant Biotechnol. J. 8 (2010) 142–154. [10] V.B. Ivleva, N.A. Schneck, D. Gollapudi, F. Arnold, J.W. Cooper, Q.P. Lei, Investigation of sequence clipping and structural heterogeneity of an HIV broadly neutralizing antibody by a comprehensive LC-MS analysis, J. Am. Soc. Mass Spectrom. (2018), https://doi.org/10.1007/s13361-018-1968-0. [11] J.L. Huang, A. Nagy, V.B. Ivleva, D. Blackstock, F. Arnold, C.X. Cai, Hydrolysiskinetic study of AEBSF, a protease inhibitor used during cell-culture processing of the HIV-1 broadly neutralizing antibody CAP256-VRC25, 26, Anal Chem 90 (2018) 4293–4296. [12] R.V. Léon Reubsaet, Characterisation of π–π interactions which determine retention of aromatic compounds in reversed-phase liquid chromatography, J. Chromatogr. A 841 (1999). [13] B.J. Ryan, G.T. Henehan, Overview of approaches to preventing and avoiding proteolysis during expression and purification of proteins, Curr Protoc Protein Sci, Chapter 5 (2013) Unit5.25. [14] P.N. Joachim Ermer, Method Validation in Pharmaceutical Analysis: A Guide to Best Practice, 2nd ed., Wiley-VCH2014. [15] I. Q2(R1), Validation of Analytical Procedures: Text and Methodology, (2005). [16] A.A. Shukla, C. Jiang, J. Ma, M. Rubacha, L. Flansburg, S.S. Lee, Demonstration of robust host cell protein clearance in biopharmaceutical downstream processes, Biotechnol. Prog. 24 (2008) 615–622. [17] X.L. Zhao, H.; Qui, J. , Reagent clearance capability of protein a chromatography: a platform strategy for elimination of process reagent clearance testing, BioProcess Int. 23 (2015) 5. [18] Y.H. Jiang, Y. Shi, Y.P. He, J. Du, R.S. Li, H.J. Shi, Z.G. Sun, J. Wang, Serine protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride (AEBSF) inhibits the rat embryo implantation in vivo and interferes with cell adhesion in vitro, Contraception 84 (2011) 642–648. [19] G.R. Mintz, A reversible serine protease inhibitor, Biopharm, DOI (1993) 34–38. [20] C. Shi, X. Zhao, X. Wang, L. Zhao, R. Andersson, Potential effects of PKC or protease inhibitors on acute pancreatitis-induced tissue injury in rats, Vasc. Pharmacol. 46 (2007) 406–411.
Table 4). After Amicon filtration, the retentate product was analyzed. AEBS-OH was not detected; therefore, AEBSF was calculated to be below 0.015 μg/mg bNAb after normalization to protein concentration. With the application of Protein A purification followed by Amicon filtration, which yielded at least 33-fold (Table 4) and 564-fold (Table 5) clearance for free AEBSF-related impurities, respectively, the combined clearance power was calculated to be > 18,000-fold (33 × 564-fold). Currently, some animal toxicity studies were reported for AEBSF [18–20], but no toxicity results of AEBSF or AEBS-OH has been stated in humans. Therefore, direct correlation for toxicity and the residual amount of AEBSF or AEBS-OH detected has yet to be established. Pending on the establishment of a human toxicity target, additional process clearance and analytical monitoring can be further developed and applied. 4. Conclusion In summary, a new method to quantify residual AEBSF-related impurities was established to support clearance monitoring. By combining a novel sample treatment step followed by RPLC-UV analysis, in-process-related impurities such as AEBSF and its derivative, AEBS-OH, could be converted to one analyte, which made the quantification simple and reliable. Method quality was also assessed and proved to be precise and accurate, which was then successfully applied for process clearance monitoring. The combination of Protein A purification with mock UF/DF approaches demonstrated above 18,000-fold removal power for AEBSF-related impurity clearance during a small scale bNAb run. Process-related impurity clearance is an essential part of any mAb development program to ensure product safety. Although a human toxicity safety level is yet to be established for AEBSF, this is the first work to describe a method to help confirm product safety related to AEBSF after a toxicity target is established. Conflict of interest The authors declare they have no actual of potential competing conflict of interest. Acknowledgment This work was supported by the intramural research program of the Vaccine Research Center (VRC), National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH). Acknowledgement to Kevin Carlton for project leadership, Kandace M. Atallah, Joe Horwitz and Jack Yang for input and support. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
23
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/jchromb
Quantification of residual AEBSF-related impurities by reversed-phase liquid chromatography☆
T
Cindy X. Cai, Nicole A. Schneck, Doug Harris, Daniel Blackstock, Vera B. Ivleva, ⁎ Kuang-Chuan Cheng, Adam Charlton, Frank J. Arnold, Jonathan W. Cooper, Q. Paula Lei Vaccine Production Program, Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Gaithersburg, MD, USA
ARTICLE INFO
ABSTRACT
Keywords: RPLC-UV AEBSF AEBS-OH Protease inhibitor Antibody Clipping Impurity Residual
During research of a broadly neutralizing antibody (bNAb) for HIV-1 infection, site-specific clipping was observed during cell culture incubation. Protease inhibitor, 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF), was supplemented to the cell culture feeding to mitigate clipping as one of the control strategies. It led to the need and development of a new assay to monitor the free AEBSF-related impurities during the purification process. In this work, a reversed-phase liquid chromatography (RPLC-UV) method was developed to measure the total concentration of AEBSF and its major degradant product, 4-(aminoethyl) benzenesulfonic acid (AEBS-OH). This quantitative approach involved hydrolysis pre-treatment to drive all AEBSF to AEBS-OH, a filtration step to remove large molecules, followed by RPLC-UV analysis. The method was qualified and shown to be capable of measuring AEBS-OH down to 0.5 μM with good accuracy and precision, which was then applied for process clearance studies. The results demonstrated that a Protein A purification step in conjunction with a mock ultrafiltration/diafiltration (UF/DF) step could remove AEBSF-related impurities below the detection level. Overall, this study is the first to provide a unique approach for monitoring the clearance of free AEBSF and its related degradant, AEBS-OH, in support of the bNAb research.
1. Introduction Monoclonal antibodies (mAbs) have been long considered for various therapeutic applications due to their capability to recognize and to provide protection against foreign invading antigens [1]. Currently, they have been applied for the treatment of a variety of diseases [2]. To support the growing needs for therapeutic mAbs, the biopharmaceutical industry has been constantly improving processes to achieve products with high yields and good quality [3]. One of the key challenges for biopharmaceutical research and development is the protein product susceptibility to proteolytic clipping during the cell culture process, leading to a negative effect on biological activity [4,5]. Various approaches have been reported to solve this issue, including eliminating the enzyme reaction site through mutation of the known proteaseclipped amino acid site(s), or minimizing the protease reactions through reduction of protease concentration or their activities [6,7].
For example, adding protease inhibitor(s) to the cell culture process can impede protease actions and result in the reduction of proteolytic clipping [8,9]. However, the protease inhibitor(s) can degrade and accumulate if added continuously during the cell culture period. Therefore, it is critical to remove the protease inhibitor(s) during the purification process. In addition, a reliable quantitative method to monitor the residual protease inhibitor-related compounds must be applied to ensure product purity and safety. In a previous study, site-specific clipping within the CDR-H3 region of a broadly neutralizing antibody (bNAb) for HIV-1 infection was discovered when a fed-batch cell culture process was employed [10]. To reduce the clipping percentage, one of the approaches during the research stage involved the application of a protease inhibitor, 4-(2aminoethyl) benzenesulfonyl fluoride (AEBSF), to the cell culture bioreactor. Besides binding to proteases to prevent proteolysis, AEBSF could hydrolyze into 4-(aminoethyl) benzenesulfonic acid (AEBS-OH)
Abbreviations: AEBSF, 4-(2-aminoethyl) benzenesulfonyl fluoride; AEBS-OH, 4-(aminoethyl) benzenesulfonic acid; TFA, trifluoroacetic acid; mAb, monoclonal antibody; bNAb, broadly neutralizing antibody; RPLC-UV, reversed-phase liquid chromatography; UV, ultraviolet/visible; CV, column volume; RSD, relative standard deviation; PBS, phosphate-buffered saline; MWCO, molecular weight cut-off:; QC, quality control; LLOQ, lower limit of quantification; UF/DF, filtration ultrafiltration and diafiltration ☆ Declaration of interest: None. ⁎ Corresponding author at: 9 West Watkins Mill Rd, Gaithersburg, MD 20878, USA. E-mail address: [email protected] (Q.P. Lei). https://doi.org/10.1016/j.jchromb.2019.03.022 Received 19 December 2018; Received in revised form 19 February 2019; Accepted 19 March 2019 Available online 20 March 2019 1570-0232/ Published by Elsevier B.V.
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
2.3. Amicon™ filtration purification To represent potential clearance of AEBSF through tangential flow filtration ultrafiltration and diafiltration (UF/DF), samples were purified using Amicon centrifuge filters with a 30 kDa MWCO. Protein A purified bNAb was spiked with 0.1 mg/mL AEBS-OH and then centrifuged using Amicon 15 mL-ultra filters for 20 min at 4000g. After each centrifugation, 7 mL of fresh diafiltration buffer was added to the Amicon filter to represent additional diafiltration volumes. This process was repeated 5 times before collecting the final retentate product. 2.4. Sample preparation for RPLC-UV analysis Scheme 1. Hydrolysis reaction of AEBSF.
All residual AEBSF were fully hydrolyzed to form one “product” peak for RPLC-UV quantitative purposes. Based on previously study, AEBS-OH stock solution was prepared by incubating 500 μM AEBSF in 100 mM Tris buffer (pH 12.0) at 37 °C for 2 h [11]. The stock standard solution was further diluted accordingly using 100 mM Tris buffer to obtain the following six concentrations: 0.5, 1, 5, 10, 15 and 20 μM. These AEBS-OH standards were subsequently used as calibrants to generate a linear standard curve. A system suitability/quality control (QC) sample was prepared by diluting AEBSF in water to a concentration of 20 μM. QC and test samples were then mixed with 200 mM Tris buffer (pH adjusted to 12.0 using sodium hydroxide) at 1:1 (v:v) ratio followed by incubation at 37 °C for 2 h to ensure complete conversion of ABESF to AEBS-OH. The final QC concentration for the analyte was 10 μM, which matched the middle point of the calibration curve. Before analysis, 200 μL of each ABESF-OH calibrant, QC and test sample were filtered through Amicon filters (10 kDa MWCO) for 30 min either at 3220 g using an Eppendorf centrifuge (model 5810r, Hauppauge, NY) or at 16162 g using a Beckman centrifuge system (model Coulter Microfuge 16, Indianapolis, Indiana), which were the highest speed settings of the centrifuges. It ensured the minimum retention volume in the filter to reduce variability caused by volume differences. This process also separated AEBS-OH (202 Da) from the large molecules prior to RPLC-UV analysis. Duplicates of each standard and sample were prepared during each run. The QC sample was injected before and after all samples as a bracketing control.
at a certain rate during the cell culture process (Scheme 1) [11]. Subsequent purification steps were then applied to remove AEBSF and its degradation product. Herein, we discuss the development of a reversedphase liquid chromatography (RPLC) method with ultraviolet/visible (UV) detection to quantify the process-related impurities, AEBSF and its degradation product, AEBS-OH. In this method, samples were first hydrolyzed, during which AEBSF was completely driven to AEBS-OH (AEBSF hydrolyzed form). The hydrolyzed sample was then filtered to separate AEBS-OH (202 Da) from larger molecules (e.g., bNAb) to minimize matrix interferences and to achieve quantification of the residual impurity. This newly developed RPLC-UV approach demonstrated good sensitivity and accuracy, which was subsequently applied to support process clearance efforts by monitoring AEBSF-related impurities at various stages of purification. 2. Materials and methods 2.1. Chemicals Chemicals were analytical grade unless specified. Acetic acid, HPLC grade water and HPLC grade acetonitrile were purchased from J. T. Baker (Center Valley, PA). Tris buffer and HPLC grade trifluoroacetic acid (TFA) were purchased from Thermo Fisher Scientific (Tempe, AZ). AEBSF hydrogen chloride was purchased from G-Bioscience (St. Louis, MO). Phosphate-buffered saline (PBS) was purchased from Hyclone (Logan, UT). Sodium hydroxide was purchased from VWR (Radnor, PA), and ethanol was purchased from Warner-Graham Company (Cockeysville, MD). The Amicon Ultra centrifugal filters (0.5 mL or 15 mL capacities) with molecular weight cut-offs (MWCOs) of 10 kDa or 30 kDa were purchased from Millipore Sigma (Burlington, MA).
2.5. RPLC-UV method set up The Acquity H-class Bio UPLC system (Waters, MA), consisting of a quaternary solvent manager, sample manager and UV detector, was operated using Empower v4.0. Chromatographic separation was performed on an AdvanceBio Peptide Mapping C18 column (Agilent, 2.7 μm, 2.1 × 250 mm), at ambient temperature and injection volume of 50 μL. The detection wavelength was set at 220 nm, which was optimized using the maximum UV-absorbance after screening in the range of 190–600 nm with UV spectrometry. Mobile phase A consisted of 0.1% (v/v) TFA in water, and mobile phase B consisted of 0.085% TFA in acetonitrile. Mobile phases were delivered at a flow rate of 0.15 mL/ min using an isocratic gradient of 100% mobile phase A for 8 min. After elution of AEBS-OH, the gradient was quickly ramped up to 95% mobile phase B and held for 10 min to wash the column, followed by column re-equilibration (100% mobile phase A) for 5 min prior to the next injection. Each run was 23 min.
2.2. Protein A purification To demonstrate clearance of AEBSF after Protein A purification, AEBSF was spiked into Protein A purified bNAb to reach the concentration around 1 mg/mL (~4000 μM) in the matrix of cell culture protein A flow-through or Tris buffer. These mock samples, as well as the cell culture harvest supplemented with AEBSF during cell culture were purified over 600 μL Protein A RoboColumns (Toyopearl AFrProtein A HC-650F) using the Tecan liquid handling platform. The system was equilibrated with 5 column volumes (CVs) of PBS before sample loading, followed by two sequential washing steps with 5 CVs solution in each. Protein was eluted at low pH and neutralized using 1 M Tris (i.e. Protein A elution). The column was cleaned and regenerated for next use. All steps were performed using a flow rate of 1.9 μL/s.
2.6. RPLC-UV method qualification The AEBS-OH calibration curve was constructed by plotting detected UV peak area versus known concentration of AEBS-OH. The analyte concentration in the test samples was determined by back-calculating sample peak areas against the calibration curve, accounting for the dilution factor. Quality characteristics of the method, including specificity, linearity, sensitivity, accuracy, precision and sample 20
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
stability were evaluated. Method specificity was assessed by evaluating matrix interference potentially caused by the Tris buffer, a Protein A purified bNAb sample, or mock cell culture harvest (cell culture flow-through spiked with bNAb but without AEBSF or AEBS-OH). Method linearity was assessed by linear regression (R2) of AEBS-OH standard curves. The accuracy of the assay was evaluated by determining the percent recovery of AEBSOH concentration detected versus the spiked at 2, 10, and 18 μM into Protein A purified bNAb samples. The method intermediate precision range was tested using two bNAb samples, pre-spiked with AEBSF (~4000 mM) in mock cell culture harvest followed by Protein A purification. Samples were analyzed by variable factors, including two analysts, two columns, and two centrifuges in three different days. Lastly, for stability analysis of hydrolyzed samples, the concentration of AEBS-OH in the QC and two bNAb Protein A purified samples was obtained at timepoints 0 h and 44 h against the standard curve created at time 0 h. The detected concentration difference was evaluated to determine sample stability after hydrolysis.
Fig. 2. RPLC-UV chromatograms (y-axis normalized) for AEBSF and AEBS-OH spiked (a) in Tris buffer (b) followed by Amicon filtration, and (c) in mock cell culture harvest followed by Protein A purification and Amicon filtration.
AEBS-OH (Fig. 2). However, the AEBSF peak co-eluted with unknown matrix interference peaks from the Protein A purified sample (Fig. 2) even after a few attempts of method optimization. It imposed a challenge on direct quantification of AEBSF, the process related impurity; therefore, an alternative approach was developed. Because AEBS-OH was well-separated from other interferences and is a naturally degraded product of AEBSF during process, a strategy to drive all AEBSF into its hydrolyzed form, AEBS-OH, was employed for the quantification study. This approach allowed measurement of a single analyte for straightforward RPLC-UV analysis and quantification. More than 99% AEBSF converted to AEBS-OH after a 2-h incubation at 37 °C within the pH range of 9.0–12.0 (Fig. S2 in supplementary). Since the pH of the Protein A purified samples typically ranged from 6.0 to 7.6 after neutralization, samples were mixed with Tris buffer at a higher pH 12.0 to ensure AEBSF was fully hydrolyzed at pH 9.0 and above. The hydrolyzed samples were then filtered to reduce large proteins to extend the column lifetime. Using this sample pretreatment scheme, detected AEBS-OH using the newly developed RPLC-UV method accounted for the total amount of AEBSF and its degradation product, AEBS-OH.
3. Results and discussion 3.1. RPLC-UV method development To reduce the site-specific proteolytic clipping of the bNAb during cell culture incubation, AEBSF was supplemented into the fed-batch cell culture media daily at the concentration of 500 μM from day 7–14 (8 days). Besides binding to proteases, AEBSF also hydrolyzed into AEBS-OH as described in a previous study [11]. A proof-of-concept experiment in the supplementary demonstrated that most of AEBSF converted to AEBS-OH, but 9.1% AEBSF stayed in its original form after a 24-h hydrolysis in cell culture-like conditions at pH 7.0 and 37 °C in Tris buffer (Fig. S1 in supplementary). This indicated that AEBSF and AEBS-OH potentially coexisted in the cell culture harvest after the last AEBSF feed (24 h before harvest) to the cell culture media. Therefore, measurement of both in-process related impurities, AEBSF and AEBSOH, was required. However, no previous quantification method has been reported for analysis of both compounds; thereby, we address the need to develop a method to quantify the total amount of AEBSF and AEBS-OH. Both compounds contain benzene ring with different functional groups, therefore, a C18 stationary phase can interact with the benzene ring, retain and separate them according to their different hydrophobicity [12]. At the initial method development stage, various C18 columns with different stationary phases were screened to retain the highly hydrophilic molecule, AEBS-OH. The AdvanceBio column was thereafter chosen since it successfully retained AEBS-OH at around 6 min and AEBSF at around 18 min (Fig. 1), which demonstrated the good separation of AEBSF and AEBS-OH. Extraneous peaks were observed after Amicon filtration but did not interfere with either AEBSF or
3.2. Assay quality assessment The specificity of the method was assessed by testing various diluents and buffers (i.e., water, Tris buffer and PBS) applied during sample handling, as well as the test sample matrices (i.e., Protein A purified bNAb sample and 100× diluted mock cell culture harvest). No matrix-related peaks were observed around the AEBS-OH elution time (Fig. S3 in supplementary). Additionally, salt in the buffer was eluted in the first 3 mins and other compounds with higher hydrophobicity eluted after 8 min (when the percentage of organic mobile phase increased) even in the most complicated matrix, such as mock cell culture harvest. Thus, no background interference was observed within the retention time of 5.5–7.5 min, providing the high specificity of the method. Linearity of the method was assessed in the range of 0.5–20 μM of AEBS-OH in Tris buffer. A representative standard curve is shown in Fig. 3. Five individually prepared AEBS-OH calibration curves all generated linear correlations of R2 > 0.99, demonstrating a good linear response. Sensitivity of the method was established based on the lower limit of quantification (LLOQ) of AEBSF standard curve at 0.5 μM (~100 ng/ mL), of which signal-to-noise (S/N) was above 10 among the 20 replicate injections. Accuracy was inferred by AEBS-OH recovery during spiking experiments. AEBSF spiked into the Protein A purified bNAb samples could potentially bind to the bNAb [13], but monitoring the modified bNAb-AEBSF was beyond the testing scope of this method. Therefore,
Fig. 1. RPLC-UV chromatograms (y-axis normalized) of (a) AEBSF and (b) AEBS-OH standards in Tris buffer (final concentration: 20 μM). 21
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
Table 3 Sample stability analysis of hydrolyzed AEBS-OH QC and Protein A purified bNAb samples at storage conditions up to 44 h. Samples
Detected AEBS-OH (μM) 0h
QC Protein A purified samples
Difference %
44 h 9.8 6.6 41.3
10.4 6.9 43.7
6.7 5.0 5.8
was able to accurately quantify the total amount of free AEBSF and AEBS-OH in solution down to 0.5 μM, with the accuracy within the acceptance precision range. 3.3. Application of the RPLC-UV method for clearance study using protein A purification
Fig. 3. A representative AEBS-OH standard curve in the detection range of 0.5–20 μM.
Protein A capture as the initial downstream purification step removed the majority of in-process related impurities. A clearance study was applied to evaluate the mass balance and clearance power of free AEBSF-related impurities, which was essential for understanding the Protein A purification removal capability and limitations [16]. Two samples were prepared (details in supplementary) for (1) a proof-ofconcept study, a purified bNAb sample spiked with AEBSF (~4000 μM), which represented the maximum amount of AEBSF added to the bioreactor during the cell culture process, and (2) the real case study, the cell culture harvest with AEBSF supplement strategy. Both samples were purified using a Protein A column. The Protein A flow-through, wash, and purified bNAb elution were all collected. The concentration of AEBSF-related impurities was detected in each fraction, as well as in the loading materials. The mass balance for the purification step was defined as input(load) = output(elution) + removal(flow-through and wash) [17], which helped track the AEBSF-related impurities at various stages in the purification system. The input was defined as the total amount of AEBSF-related compounds loaded onto the Protein A column, while the output (i.e. Protein A elution) was defined as the total amount of AEBSF-related impurities measured in the Protein A purified bNAb sample. The total mass recovery of AEBSF and AEBS-OH was calculated by (output(elution) + removal (flow-through and wash)) / input(load) × 100%. Recovery was 76.7% and 101.6% in both case studies (Table 4). The results also showed that AEBSF and AEBS-OH was removed mainly during the column flow-through step and slightly during the following wash step (Table S1 and Table S2 in supplementary), but residual amount of free AEBSF and AEBS-OH still remained in the Protein A purified bNAb sample. The normalized AEBSF and AEBS-OH amount against the bNAb amount was obtained at 1.00 and 0.123 μg per mg bNAb in two case studies (Table 4), respectively. To this end, a clearance power of AEBSF and AEBS-OH was determined by input(load) / output(elution) × 100%, which was 179-fold and 33-fold in two different case scenarios (Table 4), a proof-of concept experiment and a real cell culture harvest purification process, respectively.
Table 1 Method accuracy analysis by percent recovery of AEBS-OH spiked into Protein A purified bNAb samples. AEBS-OH (μM) Spiked
Detected
2.0 10.0 18.0
2.0 9.3 17.6
Average recovery%
RSD% (n = 4)
93.8 98.3 95.1
13.3 6.7 3.9
only AEBS-OH was spiked into the Protein A purified bNAb samples at different concentration for accuracy analysis. The recovery of AEBS-OH was between 93 and 100% with precision RSD < 10% (Table 1). Thus, good method accuracy was achieved. Precision was evaluated by analyzing the following samples. About 4000 μM of AEBSF were spiked into mock cell culture harvest in two experiments, which mimicked the maximum amount of AEBSF addition to the cell culture bioreactor. Both spiked samples were purified using a Protein A column and then tested using the newly developed RPLC-UV method. The method intermediate precision was assessed on three days (duplicate sample preparations for each sample on each run) by two different analysts, using two different columns and two different centrifuge systems. As shown in Table 2, the relative standard deviation (% RSD) was < 20% with a total of 6 runs performed. The precision was within the acceptable range for residual impurity analysis [14,15]. The stability of AEBS-OH in Tris buffer and Protein A purified samples after hydrolysis was evaluated at time-points 0 h and 44 h under a storage condition of 2–8 °C. UV peak areas of the QC sample and two Protein A purified samples were monitored at the two timepoints (0 h and 44 h), and then interpolated against the standard curve prepared at the time-point 0 h. As demonstrated in Table 3, the AEBSOH concentrations measured at time-point 0 h and 44 h had a percent difference of < 10%, which was within the acceptance range. Therefore, hydrolyzed samples stored at 2–8 °C were deemed stable and acceptable for analysis for at least 44 h, which enabled > 100 injections to be batched into a single run with high confidence. Therefore, the pre-hydrolysis step followed by RPLC-UV analysis
3.4. Application of the RPLC-UV method for clearance study using Amicon filtration UF/DF can separate small molecules from large proteins; therefore, this additional purification step has the potential to further remove AEBSF (202 Da) and AEBS-OH (204 Da) in the bNAb (about 150 kDa) product. A proof-of-concept experiment was designed at the small scale using Amicon filtration to mimic the UF/DF step. AEBS-OH was spiked into a Protein A purified bNAb sample to reach the concentration of 8.47 μg AEBS-OH per mg bNAb, which was > 8 times higher than the concentration of free AEBSF-related impurities after Protein A purification clearance study described above (1.00 and 0.123 μg/mg in
Table 2 Intermediate precision assessment of 6 runs of AEBS-OH in two separated Protein A purified bNAb samples. Protein A purified samples 1 2
AEBSF detected in protein A purified bNAb (μM)
RSD% (n = 6)
6.0 40.4
17.3 5.6
22
Journal of Chromatography B 1116 (2019) 19–23
C.X. Cai, et al.
Table 4 Clearance study results of AEBSF-related impurities for Protein A purification. Case study for AEBSF and AEBS-OH
Recovery (%)
Normalized concentration (μg/mg bNAb) in Load
(1) Proof-of-concept case study (2) Real case study
1.79 × 10 4.10
Load After filtration
1.00 0.123
179-fold 33-fold
Appendix A. Supplementary data
Table 5 Clearance study results for Amicon filtration. Samples
Protein A purified bNAb 2
76.6 101.6
Clearance power
AEBS-OH (μg/mg bNAb)
Clearance power
8.47 < 0.015
> 564-fold
Supplementary data to this article can be found online at https:// doi.org/10.1016/j.jchromb.2019.03.022. References [1] A.B. Mahmuda, F.; Al-Zihiry, K.J; Abdulhaleem, N; Majid, R.A.; Hamat, R.A.; Abdullah, W.O.; Unyah Z. , Monoclonal antibodies: a review of therapeutic applications and future prospects, Trop, J. Pharm. Res. 16 (2017) 713–722. [2] X. Geng, X. Kong, H. Hu, J. Chen, F. Yang, H. Liang, X. Chen, Y. Hu, Research and development of therapeutic mAbs: an analysis based on pipeline projects, Hum Vaccin Immunother 11 (2015) 2769–2776. [3] P. Gronemeyer, R. Ditz, J. Strube, Trends in upstream and downstream process development for antibody manufacturing, Bioengineering (Basel) 1 (2014) 188–212. [4] F. Grosse-Holz, L. Madeira, M.A. Zahid, M. Songer, J. Kourelis, M. Fesenko, S. Ninck, F. Kaschani, M. Kaiser, R.A.L. van der Hoorn, Three unrelated protease inhibitors enhance accumulation of pharmaceutical recombinant proteins in Nicotiana benthamiana, Plant Biotechnol. J. 16 (2018) 1797–1810. [5] F. Robert, H. Bierau, M. Rossi, D. Agugiaro, T. Soranzo, H. Broly, C. MitchellLogean, Degradation of an Fc-fusion recombinant protein by host cell proteases: identification of a CHO cathepsin D protease, Biotechnol. Bioeng. 104 (2009) 1132–1141. [6] H. Laux, S. Romand, S. Nuciforo, C.J. Farady, J. Tapparel, S. Buechmann-Moeller, B. Sommer, E.J. Oakeley, U. Bodendorf, Degradation of recombinant proteins by Chinese hamster ovary host cell proteases is prevented by matriptase-1 knockout, Biotechnol. Bioeng. 115 (2018) 2530–2540. [7] B.J. Ryan, G.T. Henehan, Avoiding proteolysis during protein purification, Methods Mol. Biol. 1485 (2017) 53–69. [8] S. Chakrabarti, C.J. Barrow, R.K. Kanwar, V. Ramana, J.R. Kanwar, Studies to prevent degradation of recombinant fc-fusion protein expressed in mammalian cell line and protein characterization, Int. J. Mol. Sci. 17 (2016). [9] C. Goulet, M. Benchabane, R. Anguenot, F. Brunelle, M. Khalf, D. Michaud, A companion protease inhibitor for the protection of cytosol-targeted recombinant proteins in plants, Plant Biotechnol. J. 8 (2010) 142–154. [10] V.B. Ivleva, N.A. Schneck, D. Gollapudi, F. Arnold, J.W. Cooper, Q.P. Lei, Investigation of sequence clipping and structural heterogeneity of an HIV broadly neutralizing antibody by a comprehensive LC-MS analysis, J. Am. Soc. Mass Spectrom. (2018), https://doi.org/10.1007/s13361-018-1968-0. [11] J.L. Huang, A. Nagy, V.B. Ivleva, D. Blackstock, F. Arnold, C.X. Cai, Hydrolysiskinetic study of AEBSF, a protease inhibitor used during cell-culture processing of the HIV-1 broadly neutralizing antibody CAP256-VRC25, 26, Anal Chem 90 (2018) 4293–4296. [12] R.V. Léon Reubsaet, Characterisation of π–π interactions which determine retention of aromatic compounds in reversed-phase liquid chromatography, J. Chromatogr. A 841 (1999). [13] B.J. Ryan, G.T. Henehan, Overview of approaches to preventing and avoiding proteolysis during expression and purification of proteins, Curr Protoc Protein Sci, Chapter 5 (2013) Unit5.25. [14] P.N. Joachim Ermer, Method Validation in Pharmaceutical Analysis: A Guide to Best Practice, 2nd ed., Wiley-VCH2014. [15] I. Q2(R1), Validation of Analytical Procedures: Text and Methodology, (2005). [16] A.A. Shukla, C. Jiang, J. Ma, M. Rubacha, L. Flansburg, S.S. Lee, Demonstration of robust host cell protein clearance in biopharmaceutical downstream processes, Biotechnol. Prog. 24 (2008) 615–622. [17] X.L. Zhao, H.; Qui, J. , Reagent clearance capability of protein a chromatography: a platform strategy for elimination of process reagent clearance testing, BioProcess Int. 23 (2015) 5. [18] Y.H. Jiang, Y. Shi, Y.P. He, J. Du, R.S. Li, H.J. Shi, Z.G. Sun, J. Wang, Serine protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride (AEBSF) inhibits the rat embryo implantation in vivo and interferes with cell adhesion in vitro, Contraception 84 (2011) 642–648. [19] G.R. Mintz, A reversible serine protease inhibitor, Biopharm, DOI (1993) 34–38. [20] C. Shi, X. Zhao, X. Wang, L. Zhao, R. Andersson, Potential effects of PKC or protease inhibitors on acute pancreatitis-induced tissue injury in rats, Vasc. Pharmacol. 46 (2007) 406–411.
Table 4). After Amicon filtration, the retentate product was analyzed. AEBS-OH was not detected; therefore, AEBSF was calculated to be below 0.015 μg/mg bNAb after normalization to protein concentration. With the application of Protein A purification followed by Amicon filtration, which yielded at least 33-fold (Table 4) and 564-fold (Table 5) clearance for free AEBSF-related impurities, respectively, the combined clearance power was calculated to be > 18,000-fold (33 × 564-fold). Currently, some animal toxicity studies were reported for AEBSF [18–20], but no toxicity results of AEBSF or AEBS-OH has been stated in humans. Therefore, direct correlation for toxicity and the residual amount of AEBSF or AEBS-OH detected has yet to be established. Pending on the establishment of a human toxicity target, additional process clearance and analytical monitoring can be further developed and applied. 4. Conclusion In summary, a new method to quantify residual AEBSF-related impurities was established to support clearance monitoring. By combining a novel sample treatment step followed by RPLC-UV analysis, in-process-related impurities such as AEBSF and its derivative, AEBS-OH, could be converted to one analyte, which made the quantification simple and reliable. Method quality was also assessed and proved to be precise and accurate, which was then successfully applied for process clearance monitoring. The combination of Protein A purification with mock UF/DF approaches demonstrated above 18,000-fold removal power for AEBSF-related impurity clearance during a small scale bNAb run. Process-related impurity clearance is an essential part of any mAb development program to ensure product safety. Although a human toxicity safety level is yet to be established for AEBSF, this is the first work to describe a method to help confirm product safety related to AEBSF after a toxicity target is established. Conflict of interest The authors declare they have no actual of potential competing conflict of interest. Acknowledgment This work was supported by the intramural research program of the Vaccine Research Center (VRC), National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH). Acknowledgement to Kevin Carlton for project leadership, Kandace M. Atallah, Joe Horwitz and Jack Yang for input and support. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
23