Advances in Monoclonal Antibody Purification

ABSTRACT

Challenges in the purification of monoclonal antibodies (MAbs) include reducing production cost, developing robust processes for both product purity and viral clearance, and integrating upstream and downstream processes. Increases in titers for MAb feed streams and clinical doses have led to the need to develop new products and technologies or adapt existing ones to meet these challenges. This paper will discuss the status of some of these new technologies and products and their potential for improving platform processes, as well as issues related to their implementation.

Purifying proteins for therapeutic use requires highly selective and robust technologies to achieve the very high purity required for biopharmaceuticals. Processing such valuable product streams at large-scale while manufacturing to such high standards requires careful technology design and execution. This is because feed stocks containing the target protein most often are obtained from the culture of live cells and are variable in both product content and composition. Since 2003, 26 of the 31 newly approved protein therapeutics are produced in either E. coli microbial biosynthesis or mammalian cell culture.¹ Because proteins degrade easily as a result of extremes of pH or temperature, or contact with common solvents, controling the variables during processing is especially important to avoid having to remove impurities formed during processing, because these impurities affect manufacturing yields and the activity profile of the therapeutic protein.²

(Pfizer, Inc.)

The lowest production cost is achieved not by increased process complexity but by the opposite: Operating faster and with the fewest and most efficient downstream steps reduces investments in validation and subsequent batch failures.³ Some high-dose monoclonal antibody (MAb) products require an annual production of metric tons of material to supply the market. Demand for quantities this large exceed current manufacturing capabilities requiring companies to invest in additional capability to meet demand. This delay can contribute to missed revenue. Current platforms for the large-scale production of MAbs, for example, show that it is essential to design and operate an integrated series of purification steps that are both high yielding and product molecule–specific.⁴

Emerging strategies to improve process efficiency include high-throughput technologies, using disposable equipment, and using new filters and adsorbents that combine inexpensive substrates with greater functionalities, such as charge or lipid binding. In addition, increasing volumetric titers into downstream processing also has created technological bottlenecks that must be solved. How to take advantage of the multitude of emerging technologies while streamlining and improving a platform downstream process for MAbs remains one of the major challenges faced by practitioners of the art today.

Disposables

Although some types of disposables such as media and buffer bags have been in use for years, the use of purification-related disposables is relatively new. The advantages in their implementation include reductions in preparation and product changeover, ease of implementation and validation, elimination of cleaning procedures, and a reduction in capital outlay.⁵ New and improved disposable technologies include single-use chromatography membranes and ready-to-use columns, ultrafiltration cassettes, and chromatography and ultrafiltration flow paths containing single-use monitors and transducers. However, it is not always easy to incorporate into existing processes as a one-for-one swap without additional development work. In addition, manufacturing costs must be considered, taking into account the advantages of disposables discussed above compared to disadvantages such as increased material demand and validation of leachables.⁶

The use of disposable membranes in MAb purification has shown potential to replace traditional resins, especially for anion exchange (AEX) membranes that are used in flow-through mode, in which the impurities bind to the membrane while the antibody flows through. The binding capacity of the target protein has previously been a problem for membranes because of their lower surface-to-bed volume ratio as compared to resins,⁷ so flow-through operations present an ideal situation for membrane use. The advantages of these membranes include improved mass flow properties because of the porous structure of the membrane hierarchy, and ease of preparation before use.⁸ The initial preparation time is significantly reduced because column preparation operations such as packing and qualifying, and post-use cleaning and storage are eliminated.

Figure 1 shows data comparing the DNA removal properties of two commercially available membranes, Sartobind Q (Sartorius Stedim Biotech, Germany) and Mustang Q (Pall Corporation, New York), as compared to various AEX resins. The starting material was affinity-purified broth from a Chinese hamster ovary (CHO) cell line expressing a MAb, and the flow was comparable based on a linear flow rate of 200 cm/h. The data are shown as DNA fold reduction, which is the starting value in the load material divided by the final value in the flow-through pool. The graph shows that both membranes were better at removing DNA than the resins tested, possibly because of faster, and therefore, more efficient charge interactions between DNA and the functional groups on the nonporous surface of the membrane. Additional experimentation also has shown a six-fold difference in the total amount of DNA that can be removed by a membrane as compared to resin.

Figure 1

However, host cell protein (HCP) removal by membranes can be a challenge. Side-by-side comparison of the Sartobind Q membrane to Q Sepharose Fast Flow resin has shown the membrane to be significantly less capable of removing HCP for some antibodies (Figure 2). However, new advances in salt tolerant membranes, such as the STIC from Sartorius Stedim and the Chromasorb from Millipore (Bedford, MA), which includes changes in the ligand type as well as the support and attachment chemistry, have shown significant improvements in HCP removal capability as compared with traditional membranes (Figure 2).

Figure 2

The question that remains for these disposable membranes relates to the cost of goods. Given the higher cost of membranes, their use makes sense during early-phase manufacturing when the number of batches is small, but how do they stack up in commercial manufacturing? The answer may lie in the ability of the membrane to be used multiple times. We have shown that reusing the Sartobind Q membrane up to 10 times does not affect its ability to remove DNA, but validation for reuse is an issue because it has yet to be proven.

Another advance in disposable technology is the single-use ultrafiltration cassette, such as the SIUS from Novasep (Pompey, France). In a comparison with standard multi-use cassettes, the SIUS showed a significant increase in flux capacity and reduced processing time for both concentration and diafiltration unit operations (Figure 3). The starting material in this case was a purified MAb, concentrated and diafiltered on small (100 cm² or equivalent) 50 kD cassettes using an AKTA Crossflow system from GE Healthcare (Uppsala, Sweden). Although Figure 3 shows the data for only a single antibody, the same trend was seen with other therapeutic antibodies as well. These single-use cassettes can be paired with new disposable flowpath technologies to produce a true single-use unit operation that eliminates post-use cleaning and storage. Again, we must determine the cost of goods for commercial manufacturing as opposed to early-phase use.

Figure 3

Adapting Existing Technologies

One potential purification challenge related to increased titer is the lack of capacity for Protein A resins. Several attempts to address this by removing Protein A resin from MAb purification schemes have met with not much success. However, one promising technology adapted from other industries is multicolumn countercurrent solvent gradient purification (MCSGP), which is a hybrid between simulated moving bed technology used in the chemical industry and standard batch chromatography. The adaptation of this technology for MAb purification has been explained in detail in multiple articles.^9–11

This technology can be applied either to increase the capacity of standard affinity resins in antibody purification platforms, or to increase the resolving power of non-affinity unit operations. This method potentially could facilitate the removal of affinity chromatography from an antibody purification process. Using this technology in a nonaffinity two-chromatography step process, we achieved a seven-fold increase in productivity with no reduction in impurity removal as compared with a standard affinity three-column process (data not shown). Multiple companies are in the process of developing chromatography systems and columns to optimize this technology.

A second type of technology used in other industries that has been adapted with some success is precipitation. This method can be of two types: positive precipitation, in which the product of interest is precipitated, leaving the impurities behind; and negative precipitation, in which impurities are precipitated, leaving the product of interest in the supernatant solution. Precipitation generally is used early in the purification process, to remove large quantities of impurities for increased column performance before polishing steps using resin chromatography. It has been classified as a technology with low resolution potential but high industrial maturity.¹² Chemicals such as ammonium sulfate, polyethylene glycol, and other polyelectrolytes for positive precipitation of antibodies from crude broth have been used for several years. Although they have not gained a significant foothold against standard capture chromatography methods, lately there has been renewed interest in developing them as alternatives to the standard antibody purification process in which affinity resin is the workhorse.^13,14

Recently coming into more extensive use are negative precipitants, designed to leave the product of interest in solution. One of these is caprylic acid, a short-chain fatty acid that can be added directly to conditioned media out of the reactor to remove host cell impurities such as DNA and protein. A significant reduction of DNA and HCP can be achieved using caprylic acid at concentrations of 100–500 mM and pH values of 4.0–6.0.¹⁵ However, HCP removal was achieved only for CHO-produced antibodies, not for NS0-produced antibodies.¹⁵ Even with this limitation, the use of caprylic acid has been incorporated into manufacturing processes to facilitate the use of a two-chromatography step process without Protein A resin.¹⁶

Advances in Filters

New advances in filtration, including both multilayer depth filters and nanofilters, have begun to address limitations in throughput and purification by providing new tools that increase the flexibility of order for current unit operations. These new filters also have the potential to replace standard chromatography unit operations.

A new depth filter under development by 3M Purification (St. Paul, MN), is being designed to increase flexibility in MAb purification by enhancing the removal of DNA after centrifugation, and of both DNA and HCP removal after capture-step chromatography. This filter, designated ZPQ020, is a hybrid purifier that contains both a size exclusion component and an AEX (charged) component.

Figure 4 shows the excellent DNA removal capability of this filter after centrifugation of harvest broth. The graph presents the log reduction values from five CHO-produced antibodies where a new filter was used after centrifugation. In two of the five cases, the DNA was removed completely, and in the other three, the log reduction was significant. Other filters tested showed minimal to no DNA reduction. Removing DNA before an initial Protein A capture chromatography step could potentially extend the resin life of a very expensive raw material, or even allow the resin to be eliminated altogether and substituted by a cheaper resin such as ion exchange.

Figure 4

In addition, this filter also exhibits excellent impurity removal post-capture as compared with standard AEX technologies. The results from two CHO-produced MAb tested post-affinity columns are presented in Figure 5. In both cases, these antibodies exhibited higher than normal levels of DNA and HCP. The synthetic 3M filter exhibits better HCP removal for feed streams with higher than normal impurity levels. In terms of DNA removal, the 3M filter showed better DNA and HCP reduction for the two antibodies tested than the other AEX technologies evaluated. Future studies are needed to compare this filter with the new salt tolerant membranes.

Figure 5

Advances in nanofilters also have provided flexibility in the purification process, allowing for greater throughput resulting in decreased membrane area, or reducing the processing time for comparably sized membranes. New nanofilters such as Asahi Kasei's BioEX or Millipore's VPro can significantly increase flux. Figure 6 shows the flux data for two CHO-produced antibodies comparing Asahi Kasei's Planova 20N to the new Planova BioEX. The flux for the BioEX is significantly higher than for the 20N, though with other antibodies we have seen greater flux decay for the BioEX (data not shown). The effect of this increased flux on viral clearance could be a significant concern, although initial assessments show no effect.

Figure 6

However, as we push the envelope toward higher concentrations and salt conductivities (such as in conjunction with the use of salt-tolerant membranes), we may start to see a reduction in clearance values. Figure 7 shows the log reduction values for Mouse Minute Virus (MMV) for two purified CHO-produced MAbs at a low pH and high pH value, at two higher salt concentrations (>100 mM sodium chloride). The load antibody aggregate concentration for these antibodies was also relatively high (>5%). These starting materials were filtered across a Planova 20N filter at standard operating conditions per the manufacturer's recommendations. Complete removal of virus was only seen under one set of conditions—low pH and low salt (MAb 5). For all other conditions, virus was detected in the filtrate. These data indicate that under certain conditions, the virus is extruded through the nanofilter, a factor that must be considered when designing the purification process.

Figure 7

Summary

Expression titers and dosing levels for monoclonal antibodies (MAbs) continue to increase as medical needs are successfully met with these products, placing pressure on both existing production capabilities and development decisions for downstream purification processes. This means rising production costs because extra unit operations are needed to meet the regulatory requirements for product purity and viral clearance. New technologies, such as disposable membranes and cassettes, along with new depth filters and nanofilters, or the adaptation of existing technologies from other industries, can be used to meet these production challenges. By providing flexibility not only in the order of unit operations, but in the design of the steps themselves, these new products and methods will enable purification processes that are both cost efficient and effective in producing the desired product.

Acknowledgements

The authors would like to thank the Cell Line Development and Process Development Analytics groups at Pfizer St. Louis for their support, as well as Sartorius Stedim, 3M, and Asahi Kasei for providing product samples and technical expertise.

JUDY GLYNN is a senior principal scientist, DENIS BOYLE, PhD, is an associate research fellow, JAY WEALAND is a principal scientist, ERIN MILLER-CARY is a senior associate scientist, BRIAN CHEN is a senior scientist, and PAUL MENSAH, PhD, is an associate research fellow, all at Pfizer BioTherapeutics Research and Development, St. Louis, MO, 636.247.6519, judy.k.glynn@pfizer.com

References

1. Cordoba-Rodriguez RV. Aggregates in MAbs and recombinant therapeutic proteins: a regulatory perspective. BioPharm Int 2008;21(11):44–53.

2. Fahrner RL, Knudsen HL, Basey CD, Galan W, Feuerhelm D, Vanderlaan M, et al. Industrial purification of pharmaceutical antibodies: development, operation, and validation of chromatography processes. Biotechnol Genet Eng Rev. 2001;18:301–27.

3. Kelley B. Very large scale monoclonal antibody purification: the case for conventional unit operations. Biotechnol Progr. 2007;23(5):995–1008.

4. Walsh G. Biopharmaceutical benchmarks 2006. Nat Biotechnol. 2006;24(7):769–76.

5. Low D, O'Leary R. Pujar NS. Future of antibody purification. J Chrom B, 2007;848(1):48–63.

6. Farid, S. Process economic drivers in industrial monoclonal antibody manufacture. In: Gottschalk U, editor. Process scale purification of antibodies. Hoboken, NJ: John Wiley & Sons; 2009. p. 239–62.

7. Zhou JX, Tressel T. Basic concepts in Q membrane chromatography for large-scale antibody production. Biotechnol Prog. 2006;22(2):341–9.

8. Boi C. Membrane adsorbers as purification tools for monoclonal antibody purification. J Chrom B. 2007;848(1):19–27.

9. Gottschlich N, Kasche V. Purification of monoclonal antibodies by simulated moving-bed chromatography. J Chrom A. 1997;765(2):201–6.

10. Strohlein G, Aumann L, Muller-Spath T. Morbidelli M. The multicolumn countercurrent solvent fradient purification process. BioPharm Int. 2007 Feb supplement, Advances in Process Chromatography;42–8.

11. Muller-Spath T, Aumann L, Strohlein G. Chromatographic separation of three monoclonal antibody variants using multicolumn countercurrent solvent gradient purification (MCSGP). Biotechnol Bioeng. 2008;100(6):1166–77.

12. Przybycien TM, Pujar NS, Steele LM. Alternative bioseparation operations: life beyond packed-bed chromatography. Curr Opin Biotech. 2004;15(5):469–78.

13. McDonald P, Victa C, Carter-Franklin JN, Fahrner R. Selective antibody precipitation using polyelectrolytes: A novel approach to the purification of monoclonal antibodies Biotech Bioeng. 2009;102(4):1141–51.

14. Giese, G. Polyethylene glycol precipitation of monoclonal antibodies and the impact on column chromatography. Presented at BioProcess International Conference 2009.

15. Glynn, J. Process-scale precipitation of impurities in mammalian cell culture broth. In: Gottschalk U, editor. Process scale purification of antibodies. Hoboken, NJ: John Wiley & Sons; 2009. p. 309–24.

16. Wang J, Diehl T, Aguiar D, Dai XP, Arunakumari A. BioPharm Int. 2009 Jan Supplement, Vaccine Development and Manufacturing: Pandemics and Beyond;22(10):4–10.