Advances in Non-Protein A Purification Processes for Human Monoclonal Antibodies

ABSTRACT

Cation exchange capture chromatography is the basis of the well-established non-affinity purification processes for many human monoclonal antibodies (HuMAbs). In all these processes, a concentration and diafiltration tangential flow filtration (TFF) step is essential to adjust the processing conditions for efficient capture on the cation exchange resin. Attempts to replace this preconditioning step with the precipitation of either host cell proteins (HCPs) or the antibody itself from clarified CHO cell culture broth have been successful. The application of a host cell protein exclusion strategy on the first column resulted in a significant reduction of process contaminants while maintaining high binding capacity. This allowed robust scale-up of the precipitation-based non-affinity processes. Because the demand for contaminant removal on the post-precipitation steps of the process was minimal, the antibody was purified to therapeutic quality in just two ion exchange steps. This improvement reduces the costs of antibody purification from high-titer cell culture batches still further by addressing critical constraint points during downstream processing.

Timothy Diehl

Purification processes for antibodies have improved progressively over the last 20 years so that a limited number of unit operations can now handle a more than 50-fold increase in cell culture productivity. Most current processes for therapeutic antibodies use Protein A for capture chromatography with one or more subsequent polishing steps. These affinity processes must be modified to handle increasing antibody titers. In particular, the precipitation of cell line–derived contaminants through the addition of chemicals such as caprylic acid can generate cleaner feed streams that require fewer downstream process steps.^{1, 2}

Processes that do not incorporate affinity columns for either the capture or polishing steps have also been developed successfully. These processes generally include concentration and diafiltration steps for primary recovery so that the feed stream is conditioned for efficient capture by ion exchange chromatography.³ A TFF step used for this purpose also achieves the partial clearance of process-derived DNA. There are several alternatives to replace the primary recovery TFF step, such as expanded bed chromatography⁴ or precipitation of the antibody from the cell culture broth.⁵ Selective precipitation of antibodies can be achieved by adding polyethylene glycol (PEG) for research-scale preparations, but more recently, PEG precipitation followed by two ion exchange polishing steps was recommended for commercial-scale antibody purification processes.⁵ The antibody precipitation step must be scalable to ensure consistent feed quality and efficient contaminant clearance in the rest of the process.

Here we describe the successful integration of a precipitation step into non-affinity purification processes by optimizing differential binding and elution conditions for cation exchange (CEX) chromatography resins, an approach that is referred to here as the "HCP exclusion strategy."^6,7 The comparative performance of non-affinity processes initiated with either filtration or precipitation is discussed in the context of antibody production from CHO cells at a titer of 5 g/L.

Process 1: Ion exchange process scheme with primary recovery TFF

Simple non-affinity purification processes incorporating concentration and diafiltration steps for primary recovery have been developed for many HuMAbs, and have been scaled up to work with ~100-L columns that process up to 5,000 L of cell culture broth.⁸ These processes include one or two ion exchange or mixed mode chromatography steps for polishing. The process begins with buffer-exchanged clarified cell culture supernatant from the TFF step, which is conditioned to allow antibody capture on the CEX resin. The resin can also clear most of the HCP and DNA by applying the HCP exclusion conditions. In addition, the high binding capacity of CEX resins (50–120 mg/mL of resin) can reduce the number of capture cycles. In the case of two-step non-affinity processes, an anion exchange (AEX) membrane is the only necessary polishing step. This step clears residual DNA, endotoxins, and HCP, and provides an orthogonal viral removal mechanism. The loading capacity of an AEX membrane is influenced by the purity of the process intermediate from the capture step, and is limited to a few g/mL of membrane because of the need for HCP removal (Table 1). However, in some of the three-step non-affinity processes, the Q membrane load can be increased to 20 g/mL of membrane without compromising its ability to remove adventitious viruses.

Process 1 can handle a large-scale 5 g/L cell culture harvest because of the efficient capture step, the low number of capture cycles, and the rapid processing time made possible by disposable membrane chromatography. However, the productivity of this purification process can be improved further by precipitating contaminants or the antibody as the initial step before capture chromatography.

Process 2: Ion exchange process scheme with selective contaminant precipitation

Protein precipitation is often applied in the food, blood-product, and enzyme manufacturing industries⁹ and is accomplished by the addition of salts or organic solvents, or by pH titration. Antibody manufacturers are becoming more interested in these practices because they are suitable for processing bulk products at high throughput. Precipitation technologies have the potential to reduce or eliminate the bottlenecks that will inevitably be encountered when processing antibodies at the ton scale.^10,11 They may also lead to the development of cost-effective downstream processes that require fewer unit operations for purification.

Figure 1

In the current case study, a small molecule (additive 1) was added to the clarified CHO cell culture supernatant and the pH was adjusted to precipitate most of the HCPs. The resulting precipitate was removed by depth filtration, leaving behind residual HCP in the feed. Scaling up the precipitation step is subject to variations in process parameters such as mixing speed, time, temperature, and pH, the results of which are better controlled when CEX conditions are optimized for differential binding, washing, and elution of the antibody. Transition from the precipitation step to CEX chromatography by applying the HCP exclusion strategy enables the effective integration of two-step non-affinity process schemes. Using this protocol, HCP levels can fall below 10 ng/mg after the first column (Figures 1 and 2). This degree of purity cannot be achieved in a single step when using Process 1, with diafiltration as the primary recovery step. Product quality even exceeds that achieved with Protein A as the capture step. Precipitation, combined with low pH conditioning, can prepare the feed stream for CEX chromatography with a binding capacity of up to 100 mg/mL resin, minimizing the number of process cycles while simultaneously providing virus inactivation. The material from capture is then processed by AEX membrane chromatography to remove negatively charged contaminants and adventitious viruses. Because the purity of antibodies eluting from the first column is so high, tens of grams of protein can be loaded per mL of membrane, making the AEX step faster and more economical for the application of disposable unit operations (Table 1).

Figure 2

Process 3. Ion exchange process scheme with selective antibody precipitation

Antibodies can be precipitated by the addition of salt or polyethylene glycol, and the precipitate can be separated from the bulk feed by centrifugation or filtration. In the current study, the antibody was precipitated with additive 2 and the precipitate was dissolved in a resuspension buffer with a pH and conductivity suitable for direct loading onto a CEX column. Based on the known behavior of HCPs on CEX resin, we were able to design a clearance strategy that reduced the level of HCP to 40 ng/mg of antibody. When further processed by AEX membrane chromatography, HCP levels fell below the detection limit, making a two-step non-affinity process feasible (Figure 3). Mixing, incubation time, and the temperature during precipitation must be controlled to reduce variation in the load material and to achieve consistent recovery and quality. Resuspension of the antibody precipitate requires approximately 30% of the original clarified bulk volume, therefore significantly reducing the processing time for multiple CEX cycles. The protein load on the AEX membrane is determined to ensure the robustness of the process at large scale. Dilution of the protein from the CEX column is sufficient for loading on the AEX membrane chromatography, which can be operated at very high flow rates.

Figure 3

Comparative analysis of the three processes

In Process 1, a relatively stable in-process intermediate is produced after the diafiltration step with a log reduction value (LRV) of ~3 for DNA relative to the clarified cell culture supernatant. For some antibodies, the capture step reduces HCP levels to below 100 ng/mg, allowing the development of a two-step non-affinity purification scheme. For others, HCP levels remain above 500 ng/mg after CEX and the process may require more than one polishing step. Therefore, to convert the majority of processes to two-step non-affinity schemes, either direct contaminant precipitation or selective antibody precipitation is necessary. Because the additives used for precipitation are small molecules that are generally accepted in protein purification processes, additional clearance steps are not necessary as the additives are removed in the first column's flow-through and washes.

In process 2, depth filtration can separate the bulk feed containing the target protein from the precipitated contaminants, and this bulk can be loaded directly onto the first capture column. There is no reduction in the harvest volume from the precipitation step compared to TFF, so the CEX cycle time is extended during loading. Fast-flow high-binding ion exchange resins can compensate for this increase in process time, however, especially for low-volume, high-titer processes (>25 g/L) where further dilution may not have a significant impact.¹² Precipitation does not require any specialized equipment and is easily transferable to existing large-scale manufacturing facilities. Further purification of the antibody by CEX chromatography achieves consistent product recovery and quality (Figure 2). With reference to HCP removal after the first column, process 2 was more efficient than processes 1 and 3 for many HuMAbs. Process 2 also takes the best cost advantage of disposable chromatography.

In process 3, the direct precipitation of the antibody by additive 2 (based on its hydrophobicity) also results in significant contaminant removal. However, this technique is prone to greater variation during scale-up and HCP levels are higher in CEX-purified material. Although the contaminant level was high, a single Q membrane polishing step was sufficient to reduce HCP to undetectable levels. The qualitative nature of the contaminants differs substantially between process 3 and process 1. The amount loaded onto the Q membrane is much lower than in process 2 (Table 1) to accommodate the scalability and robustness of the two-step non-affinity process. Overall, the process exploited hydrophobicity as well as CEX and AEX orthogonal separations to produce therapeutic quality material. The advantages of process 3 include volume reduction, faster processing time, and easy adaptation to facilities equipped with centrifugation or gravity filtration.

Table 1. Productivity comparison of the three non-affinity purification schemes

Protein precipitation is an efficient replacement step for primary recovery TFF and can be integrated easily into the two-step non-affinity purification process for many antibodies. The precipitation-based process can help minimize the cost of antibody therapy by increasing productivity and reducing operating costs. Similar viral clearance strategies can be implemented for both ion exchange processes with either diafiltration or precipitation of contaminants (Table 2). Process 2 has the potential to clear adventitious agents by a LRV of >4 during the precipitation step.¹³ Significant virus removal was accomplished by membrane chromatography in addition to other orthogonal viral inactivation and removal methods to produce therapeutic-grade antibodies.

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

Conclusions

A scalable and robust non-affinity process scheme with two ion exchange process steps provides a further opportunity to reduce the manufacturing costs for recombinant antibodies and thus the cost of goods. The two processing steps can be combined into an entirely membrane-based process scheme. The successful operation of contiguous anion and cation exchange membranes was recently reported for a commercial antibody manufacturing process.¹⁴ Even in the case of the slightly lower binding capacity of CEX membranes, the high flow rates can still offer further economic and operational advantages.

ALAHARI ARUNAKUMARI, PhD, is senior director of process development, JUE (MICHELLE) WANG, PhD, is assistant director, purification process development, and GISELA FERREIRA, PhD, is a senior process engineer, purification process development, all at Medarex Inc, Bloomsbury, NJ, 908.479.2451, aarunakumari@medarex.com

References

1. Glynn J. Process-scale precipitation of impurities in mammalian cell culture broth. BioPharm Int. 2008 Mar (supplement):12–18.

2. Glynn J. Development of a precipitation alternative and improvements for Protein A for impurity reduction in clarified broth containing a monoclonal antibody. BioProcess International Conference; 2008 Sept. 23–26; Anaheim, CA.

3. Wang J, Diehl T, Watkins-Fischl M, Perkins D, Aguiar D, Arunakumari, A. Optimizing the primary recovery step in non-affinity purification schemes for HuMabs. BioPharm Int. 2008 Mar (supplement): 4–10.

4. Noel R, Lihme A, Lim J, Mol O, Sinclair A. Disposable adsorption systems for monoclonal antibody processing: practical, scalable operations for multi-product facilities. Recovery of Biological Products, 2008 June 22–27; Quebec City, Canada.

5. Ramanan S, Stenson R. Method of isolating antibodies by precipitation. International patent WO/2008/100578. A2. 2008 Aug 21.

6. Arunakumari A. Extending the efficiency of cation exchange bioseparations by novel approach for high capacity Humabs manufacturing. Recovery of Biological Products; 2008 June 22–27; Quebec City, Canada.

7. Arunakumari A, Wang J. Purification of human monoclonal antibodies: non-Protein A strategies. In: Gottschalk U, editor. Process Scale Purification of Antibodies. Hoboken, NJ: John Wiley & Sons, Inc.; 2009; p. 103–123.

8. Arunakumari A, Wang J, Ferreira G. Alternatives to Protein A: Improved downstream process design for human monoclonal antibody production. Biopharm Int. 2007 Feb (supplement): 36–40.

9. Ingham KC. Precipitation of proteins with polyethylene glycol. Methods Enzymol. 1990;182:301–6.

10. Gottschalk U. (2006) The renaissance of protein purification. BioPharm Int. 2006 June (supplement): 8–9.

11. Low D, O'Leary R, Pujar NS. Future of antibody purification. J Chromatogr B. 2007;848:48–63.

12. Percivia press release: DSM and Crucell announce record achievement in PER.C6 Technology. Heerlen/Leiden, the Netherlands; 2008 June 16.

13. Seng RL, Lundblad JL. Viral inactivation process. United States patent US 4939176. 1990 Dec 20.

14. Giovannoni L, Ventani M, Gottschalk U. Antibody purification using membrane adsorbers. BioPharm Int. 2008 Dec;21(12):48–52.