Precipitation of Process-Derived Impurities in Non-Protein A Purification Schemes for Antibodies

October 2, 2009

Precipitation prior to capture chromatography offers a simple, robust, and economical method to remove CHO host cell proteins and DNA.

ABSTRACT

Precipitation of process-derived impurities prior to capture chromatography in antibody purification offers a simple, robust, and economical method to efficiently remove Chinese hamster ovary (CHO) host cell proteins and DNA. By optimizing the major process parameters—pH, caprylic acid concentration, and mixing time—and understanding their interdependency, one can develop a cost-effective process step. When precipitation is applied directly in CHO cell culture, it combines the clarification and precipitation unit operations. The direct precipitation of contaminants results in seamless transition from upstream to downstream purification processes, particularly in high cell density and high titer cell culture. As a result, demand on the purification process is significantly lowered, and a simple two-step ion exchange process is sufficient to achieve therapeutic purity.

Cation exchange (CEX) capture chromatography in non-affinity processes generally requires a feed conditioning step to lower the pH and conductivity in order to attain high binding capacity (~100 mg/mL). Typically, this is achieved by including a concentration and diafiltration step for primary recovery or, less preferably, through dilution and pH titration of the clarified cell culture bulk, resulting in large processing volumes.1

Sartorius Stedim Biotech GmbH

Primary recovery TFF offers several advantages, including batch volume reduction, partial purification of process-derived DNA, and a cleaner feed stream for better column lifetime and performance.1 On the other hand, it is a lengthy unit operation, with high costs in terms of buffer consumption and TFF cassettes. Diafiltration also adds further constraints in the case of high titers by requiring even higher buffer and tank volumes at production scale.

For those reasons, alternative technologies such as precipitation have been explored, particularly to achieve higher throughput for cell culture processes with high titers. Precipitation of the protein of interest has proven successful in food, blood, and enzyme manufacturing. For example, antibodies have been precipitated successfully at large scale by adding polymers of ethylene glycol (PEG)2 or salts3 and pH titration,4 and research-scale precipitation of contaminants by charged polymers,5 cationic detergents, or short-chain fatty acids6,7 has shown promising application at the cell culture clarified bulk (CB) stage.

In this study, we extend the application of precipitation that combines pH control and caprylic acid (a short-chain fatty acid) to remove process-derived impurities directly from CHO cell culture in the production of human monoclonal antibodies (HuMAbs).

Precipitation Method Optimization

At low pH, the hydrophobicity of the octyl moiety of caprylic acid dominates and makes acidic proteins in the solution precipitate. Antibodies with basic pIs, however, have sufficient charge to counteract that hydrophobicity and remain in the supernatant. Thus, precipitation is carried out by first adjusting the pH to the appropriate level and then adding caprylic acid while mixing the contents. In this study, effective precipitation conditions were optimized primarily with clarified CHO cell culture supernatant and then extended directly to cell culture with limited development.

To develop a robust and scalable contaminant precipitation step, three major parameters—pH, caprylic acid concentration, and mixing time—must be thoroughly studied. Caprylic acid concentration and pH are interdependent. Caprylic acid shows increasing efficiency in removing CHO host cell proteins (CHOP) and DNA as pH decreases. For example, cell culture harvest by precipitation at neutral pH does not remove sufficient levels of CHOP (Figure 1). However, controlling pH alone without adding caprylic acid does not remove significant amounts of CHOP; in the pH range of 4.0–7.0, CHOP levels were reduced by only ~20% in the absence of caprylic acid. The precipitation phenomenon exhibits two distinct phases of contaminant removal: first, a sharp (~2 log) decline of CHOP from pH 7.0 to pH 6.5, followed by steady 4-fold reduction to pH 4.0 (Figure 1).

Figure 1. Optimizing pH for the precipitation of contaminants by caprylic acid in clarified bulk (caprylic acid 1%; mixing time 1 hour)

The decision about whether to lower the pH further depends on the molecule's stability as well as the type of chromatography used in subsequent purification steps. In general, antibodies are less stable at lower pH. On the other hand, at higher pH, the binding capacity of the CEX column used in the following step will be significantly reduced. Therefore, conditions must be optimized to ensure the stability of the product and at the same time to maintain the high binding capacity of the resin. Caprylic acid concentration shows a threshold value for contaminant precipitation at any constant pH level (Figures 2a and 2b). Increasing the caprylic acid concentration from 0.1 to 0.5% at both pH 5.0 and 4.5 results in a ~2 log reduction of CHOP, followed by a negligible decline in CHOP precipitation. However, CHOP reduction was shown to be much more efficient at the lower pH level (4.5), even when the caprylic acid concentration was as low as 0.2% (Figure 2b). And at pH 4.5, product quality is not affected (see the stability discussion below). Therefore, in this case study, a caprylic acid concentration of 0.2 to 0.5% was considered effective.

Figure 2. Optimizing caprylic acid concentration for the precipitation of contaminants in clarified bulk.

Another process parameter that can influence precipitation efficiency is the length of time that the cell culture contents and the caprylic acid are mixed. Prolonging the mixing time from 30 to 120 min can significantly improve the removal of DNA and CHOP, with the longer mixing time needed particularly for the efficient removal of DNA (Figure 3).

Figure 3. Impact of mixing time on the efficiency of a precipitation step (pH 4.5, caprylic acid concentration 1%)

Overall, these three parameters (pH, caprylic acid concentration, and mixing time) must be tested interdependently. Other operating parameter ranges (e.g., temperature and mixing speed) also must be optimized for consistent performance of the precipitation step during scale-up. For the current studies, all the experiments at all scales were conducted at ambient temperature (20–25 °C).

Direct Precipitation of Contaminants from Cell Culture: Two Case Studies

Caprylic acid also has been used successfully to remove host cell contaminants directly from cell culture. In two HuMAb case studies with CHO fed-batch cultures, caprylic acid precipitation resulted in significant reductions of CHOP and DNA contaminants. In HuMAb case study 1, CHO cell culture reached a peak cell density of 12 x 106 cells/mL and a productivity of ~4 g/L. In HuMAb case study 2, cell densities were almost 10 fold (~120 x 106 cells/mL) than in the HuMAb 1 study and the titer was also much higher: 14 g/L.

The results of contaminant removal from cell culture in the first case study are shown in Figure 4. Precipitation with 1% caprylic acid with a mixing time of 1 h followed a pattern similar to that seen in clarified bulk precipitation (Figure 1). However, the only difference was a pH shift: cell culture required a pH level of 6.0 instead of pH 6.5 as in the case of clarified bulk to achieve the same level of CHOP reduction (Figures 1 and 4). This reduction also can be achieved simply by increasing the caprylic acid concentration, but the cost of caprylic acid should be taken into consideration.

Figure 4. Optimization of pH for direct precipitation of contaminants from CHO cell culture (caprylic acid concentration 1%; mixing time 1 h)

To precipitate CHOP from cell culture with a peak cell density of 120 x 106 cells/mL, 1% caprylic acid at pH 4.5 with a mixing time of 2 h was tested. The CHOP level was significantly reduced, from 3.2 x 104 to 12 ng/mg of antibody (Table 1). However, this experiment can be further optimized to define the most suitable pH level by keeping the caprylic acid level and mixing time constant. Based on the earlier results in cell culture, the right pH appears to be approximately pH 5.5.

Table 1. Scale-up of caprylic acid precipitation in clarified bulk and cell culture (pH 4.5 and 1% caprylic acid with a mixing time of 2 h)

The target CHOP value to be achieved through precipitation also can be guided by purification process capability. Optimized CEX capture for case study 1 can lower the CHOP concentration from >1,000 ng/mg to <10 ng/mg. For example, if we aim to achieve a CHOP level of ~1,000 ng/mg, a pH of 5.5 may be sufficient for HuMAb 2, whereas for HuMAb 1, pH 6.0 will be optimum (Figure 4). If precipitation is performed at relatively low pH, it is important to monitor the stability of the filtered process intermediate after depth filtration (see the section on stability below).

Post Precipitation Filtration

Caprylic acid precipitates form three layers after centrifugation. The top layer is a mixture of residual caprylic acid and the CHOP precipitate (Figure 5), and the bottom layer, the sediment, is enriched with precipitated DNA. Therefore, centrifugation may not be a practical option at large scale, in which case optimized depth filtration may be necessary.

Figure 5. Caprylic acid precipitation of contaminants from clarified bulk

Although depth filtration is commonly used to remove the biomass or contaminant precipitate, high cell density cell cultures pose several challenges for depth filtration. The significant costs involved in building filtration trains with multiple pre-filters makes a clarification step much more expensive than using a simple filtration assembly. By precipitating contaminants directly in cell culture, we can reduce costs by combining two processing steps for the recovery of product. In other words, the clarification of cells and removal of caprylic acid precipitate can be achieved by one depth filtration step. The removal of contaminant precipitates and scale-up are crucial for making caprylic acid precipitation a feasible procedure for large-scale manufacturing. In this study, caprylic acid precipitation has been successfully scaled up 2,000-fold in clarified bulk procedure and 200-fold in cell culture (Table 1).

In a high density cell culture with continuous mixing mode with 1% caprylic acid, filtration capacity can reach ~80 L/m2. When the settlement of the precipitate is performed before filtration, capacity can be increased to as high as ~150 L/m2, which is almost as high as the filtration capacity of cell culture clarification. Another alternative is to add a filter aid to the mixture after precipitation is completed, followed by depth filtration. Filter aids such as Celpure (Advanced Minerals, Santa Barbra, CA) can help increase flux and filter capacity by separating the solids in the feed8 and increasing permeability. Using a filter aid before depth filtration for high solid contents, as in the case of caprylic acid precipitate, improves the quality of the feed stream as measured by turbidity units. The filtration capacity can exceed 1,000 L/m2 for clarified bulk or ~500 L/m2 for bioreactor cell culture contents after precipitation with the addition of Celpure. Provided there is enough space to hold solids, filtration capacity is dependent on the filter pore size, filter aid amount, flow rate, and pressure.

Integrating Precipitation into Non-Affinity Purification Schemes

Post caprylic acid precipitation, the filtered bulk is suitable to load onto either an affinity column or a non-affinity column. Here, we describe the integration of precipitation into a non-affinity purification scheme with CEX as the first chromatography step.

The low pH condition after the caprylic acid precipitation procedure prepares the loading condition for high binding (~100 mg/mL) onto CEX resins. The levels of process-derived impurities have already been significantly reduced after precipitation under optimal conditions (Figure 6), leading to a much cleaner feed stream for the capture resin. As a result, the demand on the purification process is low enough that purification can be achieved by only two orthogonal separation steps. Typical step recoveries for precipitation are in the range of 80–99% for various HuMAb processes.

Figure 6. Product recovery, CHOP, and DNA levels following precipitation and CEX chromatography

Impurities are at very low levels after CEX capture (<10 ng/mg HCP), making it possible to complete the purification process by adding just an anion exchange (AEX) membrane chromatography step (Figure 7a). The AEX chromatography in this case is mainly for adventitious virus removal; a 3-log reduction of A-MuLV was still observed even at a conductivity level of 12 mS/cm (Figure 8), thereby accommodating the high conductivity CEX eluate with very low or no dilution.

Figure 7. Comparison of the tangential flow filtration (TFF)-based and precipitation-based process schemes

With the most recent improvement in membrane chromatography, salt-tolerant membrane adsorbers have been developed to accommodate a wide range of salt concentrations in the load while still maintaining contaminant clearance and high flow rate characteristics. Using salt-tolerant interaction chromatography (STIC) disposable membrane modules (Sartorius Stedim Biotech, Goettingen, Germany), the process can be developed without any further dilution after CEX chromatography. STIC membrane chromatography was evaluated here for contaminant clearance to replace the AEX membrane step. The STIC membrane was able to significantly reduce CHOP from CEX eluates in a polishing step without any further dilutions (Figure 9). STIC can replace a Q membrane because the viral clearance capacity was reported to be 4–5 LRVs using the model bacteriophage virus ΦX174, especially at a high salt concentration where Q membranes have negligible clearance.9 It is clear that in a precipitation-based process where CHOP levels are extremely low after CEX, STIC may be able to process loads as high as those that Q membrane (20 g/mL) can when developed for viral removal only. Further studies are needed to confirm the scalability of STIC and its robustness to handle various buffer matrixes used in non-affinity processes.

Figure 8. Effect of salt concentration on viral clearance by Sartobind Q at 20 g/mL load

The residual caprylic acid concentration in the product was monitored for clearance during the purification process. In an experiment with 1% caprylic acid addition after depth filtration, only ~0.1% residual caprylic acid was present, which was further reduced to 0.003% after CEX chromatography alone. Several other fatty acids that are constituents of cell culture medium also were removed in a CEX flow-through fraction. CEX chromatography is a crucial purification step to remove cell cuture additives to significantly low levels in a two step purification scheme.

Figure 9. CHO protein (CHOP) removal profile for a salt-tolerant membrane used in flow-through mode (pH 7.0; conductivity 12 mS/cm).

The precipitation step also significantly contributes to the viral removal strategy of non–Protein A purification schemes by adding ~4 LRV for A-MuLV.10 In our HuMAb case study 1, at pH 4.5 and with a mixing time of 2 h, 1% caprylic acid treatment provided a 4-log reduction in A-MuLV (Table 2). These studies will be extended for a range of model viruses in the future. The process capability of TFF-based and precipitation-based non-affinity purification processes to remove viruses is comparable when the CEX and Q membrane steps are performed under the same conditions for both process schemes (Table 2, Figures 7a and 7b).

Table 2. Viral clearance comparison for two non-affinity purification schemes

The economics of two different HuMAb production processes, one using a TFF-based three-step non-affinity purification scheme and the other using a precipitation-based two-step process, were compared (Figures 7a and 7b) by modeling the processes with SuperPro Designer (Intelligen, Scotch Plains, NJ). The two models were compared for raw materials, consumables, and labor costs. After analyzing the processing time, overall process recovery, and cost (2 g/L, 5,000 L compared to 10 g/L, 2,000 L), it was obvious that the precipitation process scheme considerably lowers overall costs as a result of the efficiency of the precipitation step and because fewer processing steps are needed (Figure 10). As the product titer increases, these trends become more obvious, indicating that precipitation is favorable for high titer processes.

Figure 10. Economic analysis of precipitation-based non-affinity purification processes

Product Characterization and Stability

The quality of the product after caprylic acid precipitation and purification by CEX was compared with a reference standard that was purified by a TFF-based non-affinity process. The results of isoelectric focusing (IEF) gel analysis are shown in Figure 11. Furthermore, SDS–PAGE analysis, monomer purity assessment by SEC–HPLC, and activity by binding ELISA, are similar for the two products. The stability of caprylic acid treated filtrate at 2–8 °C was also monitored. The results of a case study showed that the filtrate can be stored at 2–8 °C for 34 d without any loss of activity (102% on day 50 of storage) and monomer purity. These stability studies will be continued on the final purified antibody to reveal the effect of caprylic acid precipitation process on long-term storage.

Figure 11. IEF analysis of stability samples following caprylic acid precipitation, stored at 2–8 °C

Conclusions

The potential application of already successful and proven alternative technologies from other areas has been explored for antibody purification.11 High throughput process steps that are desirable for handling escalating titers will find their place in future large-scale manufacturing processes. Precipitation is one such alternative, and when applied to either the product or contaminants12 it significantly reduces demand on downstream steps. In addition, the precipitation of contaminants directly in the cell culture offers the further advantage of reducing unit operations at large-volume stages of the recovery processes. Because the precipitation step for cell culture can be performed directly in the bioreactor, using a disposable bioreactor is very convenient and lessens the burden of cleaning validation. Clearly, for a generic two-step non-affinity process, we have demonstrated a significant reduction in the cost of goods as well as improvements in productivity for a model HuMAb manufacturing process.12 However, economic comparisons need to be extended to various processes to account for different purification schemes and manufacturing scales used for each molecule.

JUE (MICHELLE) WANG, PhD, is the assistant director of purification process development, TIMOTHY DIEHL and DEENA AGUIAR are scientists in purification process development, XIAO-PING DAI, PhD, is the assistant director of bioprocess development, and ALAHARI ARUNAKUMARI is the senior director of process development,.all at Medarex, Inc., 908.479.2451, aarunakumari@medarex.com

References

1. Wang J, Diehl T, Watkins-Fischl M, Perkins D, Aguiar D, and Arunakumari A. Optimizing the primary recovery step in non-affinity purification schemes for HuMabs. BioPharm International, 2008, March supplement: 4-10.

2. Sundar, R. and R. Stenson. (2008). Method of Isolating Antibodies by Precipitation. WO 2008/100578 A2.

3. Habeeb A, Francis R. Preparation of human immunoglobulin by ammonium sulfate precipitation. Vox Sang. 1976;31:423-434.

4. Thommes, Gottschalk U. Alternatives to packed-bed chromatography for antibody extraction and purification. A book Chapter in Process Scale Purification of Antibodies. Edited by U. Gottschalk. Published by John Wiley & Son, Inc. (2009)

5. Clark K, Glatz C. Polymer dosage considerations in polyelectrolyte precipitation of protein. Biotechnol Prog. 1987;3(4):241-247.

6. Glynn J. Process-scale precipitation of impurities in mammalian cell culture broth. BioPharm International, 2008, March supplment: 12-18.

7. Glynn J. Process-scale precipitation of impurities in mammalian cell culture broth. A book Chapter in Process Scale Purification of Antibodies. ;Edited by U. Gottschalk. Published by John Wiley & Son, Inc. (2009)

8. Russell E, Wang A, Rathore A.S. Harvest of a therapeutic protein product from high cell density fermentation broths: principles and case study. A book chapter in Process scale bioseparations for the biopharmaceutical industry edited by Shukla A., Etzel M, Gadam S. published by CRC Press, Taylor & Francis group (2007).

9. Faber R. Novel membrane adsorbers for large scale downstream processing. 6th HIC/RPC Bioseparation conference, Napa, March 15-19, 2009.

10. Seng R.L., Lundblad J.L. Viral inactivation process. US patent number 4939176, July 3, 1990.

11. Gottschalk, U. The renaissance of protein purification. BioPharm International, 2006, June Supplement: 8-9.

12. Arunakumari, A., Wang J. and G. Ferreira. Advances in non-protein A purification processes for Human Monoclonal Antibodies. Biopharm International, 2009, March supplement: 22-26.