OR WAIT 15 SECS
This flexible setup minimizes the number of purification process steps, buffers, and process components.
A purification process scheme was developed to purify therapeutic-grade human monoclonal antibodies. It features two columns with interchangeable chromatography steps and non-affinity resins. The purification scheme includes a cation exchange column (CEX) and a hydrophobic charge induction (HCI) column and eliminates the need of an in-process diafiltration step. Viral clearance from this scheme is efficient due to the ability of HCI resin to remove adventitious agents. The scheme was scaled up 1,000- to 10,000-fold with an average overall yield >70% for a variety of antibodies. Contaminant removal and product quality from this process are comparable to those of three-column affinity and non-affinity purification schemes.
The economics of large-scale purification of proteins is important, especially for therapeutic antibodies. Antibodies make up a large percentage of the therapeutic biologics on the market, accounting for about 30% of total recombinant protein product sales. Revenues for 20 approved therapeutic antibodies are estimated to be about $17 billion in 2006, and expected to be over $22 billion in 2007.1 Costs associated with antibody-based therapies tend to be particularly high because these are expensive molecules to develop, produce, and often require high doses.
The purification scheme is very important for both final product quality and process economics, because chromatography alone can account for two thirds of downstream processing costs. When the products are monoclonal antibodies, the resin cost for an affinity-capture column such as Protein A can overwhelm the raw materials costs.2
Affinity chromatography often is used as a capture step to meet purity, yield, and throughput requirements, despite the development of advanced chromatography resins with improved performance at high flow rates and binding capacity. Traditionally, the affinity-based capture step (Protein A or G) is followed by at least one or two other chromatography steps. In addition, one or more in-process diafiltration steps are needed to condition the feed into the following column. Not only is Protein A resin at least four to five times more expensive than non-affinity media, it can also have issues such as ligand leaching. In general, even if affinity chromatography is used, adequate purity and viral clearance often are not achieved unless one or more polishing steps are interspersed.
By eliminating specific steps in downstream processing, productivity can be maximized while maintaining the integrity and purity of the molecule. We describe here a flexible, interchangeable, non-affinity, two-step purification method that can be implemented at lower cost compared to affinity-based schemes due to its simplicity and short number of steps involved. Overall recovery is high and final product quality is equivalent to more traditional protocols.
Not only did we reduce the number of steps with the proposed purification process, but we also eliminate the need for an in-process concentration or diafiltration (which would mean more development, optimization, scale-up, cleaning and cleaning validation, and increased process cost while risking potential loss or modification of product). Minimizing the number of steps will cut down the number of process components, buffers, tanks, and miscellaneous equipment. Finally, the two-column purification process can be readily implemented into the clinical manufacturing of many antibody molecules with less initial investment on development cost, time, and resource requirements.
This purification process explores the integration of the separation principles of two different resins, a cation exchange (CEX) resin (Poros 50HS or Fractogel EMD SE HiCap) and a hydrophobic charge induction (HCI) resin (MEP HyperCel). The proposed scheme adapts to different feed types and user implementation choices, while still clearing host cell proteins (HCPs), viral-like particles, nucleic acids, product-related contaminants, and media additives (e.g., methatrexate, insulin, vitamins) to levels that allow the therapeutic use of the final purified material.
This process is highly flexible because the material can flow either way. The CEX and the HCI resins can be used for either capture or a polish step, without any need for an in-process tangential flow filtration (TFF). Using CEX chromatography for capture takes advantage of the higher pI of human antibodies (typically higher than 8), which allows their capture or binding at a pH close to neutral (Ex. 6.2). The HCI resin can accommodate a wide range of conductivity values, provided that the pH of the load is maintained close to or above neutral. The viral inactivation step can be implemented between chromatography steps or after the second column. The convenient place to carry out viral filtration is at the end of the process, before product formulation, when volumes are smaller, more manageable, and the product is fully purified. This method has proven to be scalable (from about 1 mL to 10,000 mL column size, when CEX was used for capture) with reproducible results for different human antibodies.
The unique properties of MEP HyperCel (Pall Corp., East Hills, NY) make process simplification possible. MEP HyperCel, a HCI-based resin is dependent on binding pH (neutral and above) but is independent of the conductivity of the load (up to 1 M NaCl equivalent). This is what makes it extremely flexible. When MEP HyperCel is used for capture, the resin is insensitive to the high conductivity of the cell culture supernatant, allowing straight loading of the harvest. Pre-conditioning the feed during the recovery TFF can extend the life of the resin as a capture step.
The tolerance of MEP HyperCel to a wide range of conductivity during sample load is particularly useful when fed a neutralized CEX eluate. Consequently, this resin allowed for interchangeability in the proposed purification scheme because it can be applied either to capture or polish.
An additional benefit provided by this resin is seen during elution. As the pH is reduced, product is recovered due to electrostatic repulsion between the ligand and the antibody, and conductivity has no significant influence in this elution process. Therefore, when MEP HyperCel resin is used for capture, the eluate can be collected at low conductivity to allow for effective binding onto the CEX column. Consequently, product can flow directly from the HCI column to the CEX column without needing any in-process TFF. As a polishing step, there is more flexibility in the conductivity of the elution buffer, because HCI is now the final chromatography step. In addition, MEP Hypercel can be viewed as both an alternative to TFF and a chromatography step. An interesting option would be to use the formulation buffer when HCI is being used as a polish.
Several studies were performed on the HCI protocol to minimize HCPs during elution. Decreasing the conductivity during elution and increasing the elution pH from 3–4 to 4.8–5.2 favored a preferential elution of the product with a minimal relative amount of HCPs during a capture chromatography study (Figure 1). Also, we achieved higher purity by decreasing the elution flow rate (Figure 2). Other parameters such as elution at low conductivity and post-load washes with phosphate buffers also helped to exclude HCP from the purified antibody.
Finally, we observed that the binding capacity of this resin can be manipulated not only by the product's residence time in the column, but also by manipulating buffer species during equilibration and load. For example, when the strength of phosphate-based buffers was reduced from 35 mM to 10 mM, binding capacity was increased by about 170%, from 13 mg/mL to 22 mg/mL, for a specific antibody.
Conditioning cell harvest
Cell culture supernatant containing fully human monoclonal antibodies was concentrated and diafiltered in either 35 or 70 mM sodium phosphate (pH 6.2). For experiments in which MEP HyperCel was used as the capture resin, the load was adjusted to neutral pH prior to load.
Antibodies were loaded onto CEX resins with a binding capacity ranging from 15 to 45 mg/mL. However, these capacity values can be further increased to over 100 mg/mL with higher binding resins. The product was step-eluted at pH 6.2 with added salt, ranging from 40 to 75 mM NaCl. MEP HyperCel (Pall Corp., East Hills, NY) was the HCI resin. Equilibration used 10, 35, or 70 mM sodium phosphate (pH 7.0 to 7.2) and the product was loaded to a capacity of up to 22 mg/mL. The resin was post-load washed with 10 to 45 mM sodium phosphate (pH 6.2 to 7.0) and then washed before elution with a low conductivity buffer. Elution took place when buffers with a pH range of 4.5–5.2 were used, using another low conductivity buffer such as 10 mM sodium acetate or a dual acetate and phosphate buffer. A summary scheme of the proposed method can be seen in Figure 3.
Viral inactivation, filtration, and clearance studies
Low pH hold was performed for one hour. Final viral filtration used Planova 20N (Asahi Kasei Pharma, Westbury, NY). Viral clearance studies were performed by an independent laboratory (Apptec, St. Paul, MN).
Host cell protein and DNA specific concentration
Chinese hamster ovary-host cell protein (CHO HCP) analysis used Cygnus Technologies ELISA Kit (Catalog# F015). DNA was quantified by a proprietary qPCR assay.
Both versions A and B of the purification process yielded comparable quality in the final purified material. With similar materials and the two interchangeable sequences, we achieved a final homogeneous composition containing less than 100 ppm in CHO HCP, 10 pg DNA/mg antibody, and a monomer percentage >95%, measured by size exlcusion HPLC.
Table 1A summarizes the results obtained for A scheme when CEX chromatography is used for capture. This "direction" was adopted with four different antibodies. Host cell protein content was reduced 50- to 200-fold after the first column, with concentrations ranging from about 900 to 6,000 ng/mg, which were reduced to levels less than 100 ng/mg after the HCI column. The CEX resin cleared DNA very efficiently (as much as 7,000-fold), and remaining traces were further reduced by the HCI resin to less than 10 pg DNA/mg antibody.
Table 1A and 1B. Summary of in-process contaminant clearance. In Table 1A, a cation exchange (CEX) resin is used for capture, and hydrophobic charge-induced (HCI) resin is used for polishing. In Table 1B the resins were reversed.
The reverse scheme (B) yielded a comparable quality of material (Table 1B). HCP was reduced about 90-fold by the HCI. DNA was cleared much more efficiently, when compared with the CEX resin as capture, bringing the measurements below the set final specification, which emphasizes the efficiency of MEP HyperCel in the clearance of nucleic acids.
Final purity (% monomer) and overall recovery for both HuMAb-1 and HuMAb-2 purification schemes, in both directions, were extremely comparable: over 99% and approximately 70%, respectively.
Table 2 summarizes the scalability of the proposed process for two other human monoclonal antibodies. The largest variability was observed with the DNA measurements. All remaining parameters were very comparable. Final purified material retained the quality obtained at bench-scale. Recovery values tended to be better with scale-up.
Table 2. Comparison of in-process contaminant clearance using the cation exchange (CEX) resin for capture and a hydrophobic charge induction (HCI) resin (MEP HyperCel) for polish as the process is scaled-up. The overall yield for HuMAb-6 and HuMAb-7 for large-scale process was about 80% for both antibodies.
Regulatory agencies require at least two separate steps for viral inactivation and clearance during the production of therapeutic proteins. These steps must be based on different modes of action, typically, a low pH hold and a viral filtration step.3 Viral inactivation was conveniently positioned between the chromatography steps, but this step can also be positioned at the end of both sequences. Similarly, an additional specific step, such as viral clearance filtration, is performed after the second column.
Table 3 shows the viral clearance of HuMAb-6 and HuMAb-7 following scheme A. Quantification of virus-like particles in the cell culture supernatant yielded up to 10 logs of infectious and non-infectious viral particles. The minimum safety factor for clearance of A-MuLV for the process ranged between 7 and 10 logs, for HuMAb-6 and HuMAb-7, respectively. Therefore, the final antibody product (antibody and residual process contaminants) had sufficient viral clearance and safety viral factor, allowing its use as a therapeutic.
Table 3. A-MuLV clearance capacity of the purification process using a cation exchange resin for capture and hydrophobic charge induction resin for polish. Results were estimated from bench-scale runs.
With increasing titers in antibody manufacturing, purification processes must increase process efficiency and productivity. By minimizing the number of purification process steps, the number of buffers to be prepared and process components are also reduced. The two-column purification technology design described here can be used with various types of proteins with only minor process adjustments. The scheme is a starting point for subsequent optimization and iteration steps because chromatography parameters often can be optimized, for example, for higher binding capacity ranges. One additional advantage of using this scheme is the flexibility of choosing the direction of the process steps that is most convenient for a specific product or production facility.
Gisela M. Ferreira, PhD, is a process engineer, and Jill Dembecki and Krina Patel are associate scientists. Alahari Arunakumari, PhD, is the director of process development, all at Medarex, Inc., 908.479.2451, firstname.lastname@example.org
1. Rathin D and Morrow, KJ Jr. Progress in antibody therapeutics. Am Biotechnol Lab. 2006; 24(8):8–10.
2. Thompson R. Antibody therapeutics: product development, market trends, and strategic issues. 2006 Oct. Available from: www.zangani.com/blog/antibody-therapeutics-product-development-market-trends-and-strategic-issues
3. Zoon KC. Points to consider in the manufacturing and testing of monoclonal antibody products for human use. US FDA, CBER. Rockville, MD; 1997 Feb 28.