Comparing Alternative mAb Purification Tech

By Jack Vicalvi, principal scientist, J2 Biopharm Associates LLC

We have seen more improvements in downstream process development over the past 15 years than had occurred in the previous 35 years. This is the first of a two-part overview of selected innovative products and approaches with universal implications that have contributed enhancements allowing the manufacture of large molecule biological therapeutics, such as monoclonal antibodies (mAbs) and viral vectors, with higher recoveries and purity.

In Part 1 we examine how these products change the way we manufacture mAbs, where they can be used in current process development schemes, and what advantages they confer over previous methods. We touch upon the situational economics that drive the use of these innovative products or approaches. Finally, we focus on GMP scale-up models for practical purposes: if it doesn’t scale, it doesn’t matter how good it may be.

mAb Process Flow Diagrams

Clarification of the bioreactor harvest is the first step in the downstream process; it is not the last step of the upstream process since it is essentially a purification of drug substance. This step is critical for efficient purification of the product of interest; simple size filtration of the harvested material is the traditional method utilized. However, depending upon the critical quality attributes (CQAs) required, size-based filtration alone may no longer be adequate or unavoidable.

The use of charged filtration devices (such as the 3M SP and ZB filters or Millipore DOHC, COHC, and XOHC membranes) can result in up-front removal of highly charged contaminants, such as host cell DNA (hcDNA) and some host cell proteins (HCP) and viruses, significantly improving downstream capture efficiency and, therefore, initial recovery and purity. In situations where cell viability is low (<50%) during harvest, there is a marked increase in host cell debris, including DNA, HCP (including proteolytic enzymes), and intracellular or endogenous viral particles where an additional reduction via charged membrane filtration may become significant during subsequent purification steps. In situations involving low cell viability, the addition of an endonuclease, preferably in the bioreactor, for 1 hour prior to harvest would serve to reduce DNA strand length and improve filterability significantly. The endonuclease continues to work during clarification up to capture (usually 3 to 8 hours, depending upon capture conditions and loading speed).

In most cases, mAbs do not bind to anionic matrices; however, it may be necessary to adjust the pH and/or conductivity of the harvest (again, preferably in the bioreactor), depending upon the target of interest, to prevent binding to the filter or to prepare the target for capture in subsequent steps. The conditions for optimal recovery of product with maximal reduction of contaminants will have to be determined empirically for each mAb.

When appropriately sized depth filters are selected, centrifugation becomes redundant at best; then there is the problem of scale-up and cleaning validation issues, even with disposable buckets (capital equipment costs and maintenance, aerosol generation, and disposition of used buckets). In those cases where filtration is impractical, the use of expanded bed chromatography could be a more feasible and more economical option (see Upfront’s Rhobust EBA technology).

Typically, the capture step will employ a Protein A resin. However, not all Protein A moieties are created equal. Some bind different mAbs with varying degrees of specificity and avidity; therefore, any Protein A resin you choose should be determined empirically. Also, there is the resin matrix to consider; these include soft resins such as agarose and some methacrylates and the rigid resins such as polystyrene or ceramic. All have extensive regulatory files.

The primary differences between soft and rigid resins are in linear velocity and buffer consumption during operation. Agarose is a soft resin requiring wall support or the bed sags, forming a “smile” that produces a leading edge during elution in columns wider than 5 cm. It also is a slow diffusive medium and loses capacity and bed integrity above 300 cm/hr. In addition, due to its diffusive nature, it requires extensive washing during buffer changes, especially prior to elution. Capto resins are cross-linked agarose, which may alleviate some of these problems; however, the base matrix is still agarose, requiring extensive washing and slow flow rates to allow diffusion. Rigid resins are easier to pack, require much less washing, and maintain bed integrity without wall support. These resins also may be operated at linear velocities of 800 to 1,200 cm/hr without significant loss of productivity.

An inexpensive alternative to Protein A capture is CIEX. Following an adjustment to pH 6 in the bioreactor prior to clarification, most mAbs may be bound using a cation exchange resin, such as POROS HS or Toyopearl GigaCap CM. The CIEX resins also have much greater capacity than Protein A resins (although the selectivity is reduced). Rigid CIEX resins are capable of significant virus removal when operated at 450 to 600 cm/hr (3 to 4 log removal of virus, or LRV). Following the selective elution of the target, the eluate may be adjusted down for a low pH hold step, as in the typical Protein A step.

Another alternative capture method would be Toyopearl MX-TRP-650M, which utilizes tryptophan as the capture ligand. Other options such as Protein G and “boutique” ligands are significantly more costly than Protein A, posing greater risk and expense, especially in GMP manufacturing scenarios.

Table1. mAb Capture Step Comparison of Soft vs. Rigid Resin vs. CIEX (rigid resin)

*Mab SelectSure @ $16,000/L = $380,000/column x 2 (GMP back up) = $760,000 + columns ($40,000-$60,000 each).**Mab Capture A @ $10,000/L = $250,000/column x 2 (GMP back up) = $500,000 + columns ($40,000-$60,000 each).***Toyopearl GigaCap CM @ $6,000/L = $150,000/column x 2 (GMP back up) = $300,000 + columns ($40,000-$60,000 each).

The traditional approach here is to use CIEX. The same considerations outlined above would hold here as well. In the alternative approach where CIEX is used as the capture step, the initial mAb polish step is usually HIC, such as a phenyl or butyl resin. The most popular are the methacrylate-based resins from Tosoh. These resins can be operated at higher linear velocities than the agarose-based resins, generally 300 to 600 cm/hr. A mixed modal resin, such as CaptoAdhere, Capto MMC, or Hydroxy Apatite could be used at this step; these resins also may provide significant contributions to viral clearance, as well.

HIC columns are usually loaded with the CIEX eluate at conductivity from 30 to 50 mS/cm (250 to 400 mM NaCl) and subsequently washed, eluted, and stripped with decreasing conductivity steps. One of the differences between agarose and methacrylate resins is the tendency of agarose to contract in the presence of higher salts and expand in the absence of salt; methacryaltes and rigid resins do not. Elution from the HIC at physiological salt can be manipulated by increasing the pH, as well. This sets up the next step in the process.

The tendency here is to use an AIEX column. This step is operated as a flow-through for the mAb, while removing the residual contaminants remaining after capture and initial polish steps, including HCP and DNA, and it is important as a major virus removal step. Usually, the remaining contaminants are very low and, depending upon the level of residual contaminants, the column may be replaced with a membrane device.

At this point in the process the remaining contaminants are usually in µg levels; this allows the use of a rigid resin in a short bed height column (8 to 12 cm) that can be operated at high linear velocity (450 to 600 cm/hr). An alternative could be Sartobind Q or Emphaze AEX membrane devices. The key advantage of the Emphaze AEX is its virus removal capacity (4 to 7 LRV), which exceeds the resin LRV capacity (3 to 5 LRV).

The main players here are the Millipore ViResolve and Asahi-Kasei Planova 20N and BioEX. The Millipore filter is plate-and-frame, whereas the Asahi-Kasei is a hollow fiber module. The choice is dependent upon the ultimate scale for manufacturing. If the manufacturing process scale remains below 5,000 L, we prefer the Asahi-Kasei BioEX or Planova 20N since there is no prefilter required and the device is easier to handle as it is self-contained (no holder required). At higher manufacturing volumes, the Millipore ViResolve may be more appropriate even though a holder and prefilter are required.

In traditional TFF, filters are used in a process and then cleaned between uses with a clean-in-place (CIP) operation. This requires a significant investment in system capacity, raw materials, energy, and time. There are three main players here: Millipore, Pall, and Repligen. Both Millipore and Pall produce multi-use membranes requiring cleaning and storage validation steps. The only disposable, single-use device is the Repligen TangenX line. Repligen’s TangenX SIUS Single-use TFF Cassettes perform like reusable cassettes with single-use convenience at a fraction of the cost. These are a convenient alternative delivered ready-to-use with a range of pore sizes and flow-channel configurations validated for GMP use. They ship pre-sanitized and packaged with 0.2 M NaOH. These cassettes are compatible with holders from Millipore, Pall, and Sartorius; newer models come packaged in disposable holders.

An alternative approach is Pall’s Cadence Single-Pass TFF (SPTFF), a patented breakthrough technology that allows direct flow-through concentration with no recirculation of product. SPTFF enables high concentration of shear sensitive proteins and antibodies >160 grams/liter. What makes this possible is a unique flow path design and staging of cassettes in series. Volume capacities range from several liters to thousands of liters. The Cadence SPTFF process is continuous, producing high-concentration factors and eliminating the need for the conventional recirculation loop, which minimizes aggregation problems and requires no mixing, minimizes shear exposure, and allows the SPTFF step to be coupled with other downstream process steps. Additional SPTFF benefits over conventional TFF include lower system hold-up volumes, higher recoveries, and lower flush volume requirements. However, the operating conditions and module configuration for any single-pass TFF process must be established by empirical testing.

Since the harvest contains proteolytic enzymes and cell debris, these must be removed as early as practicable. The important thing is to start the process with your best chance to not only capture your target molecule but to remove as many contaminants as possible.

At this stage, we should talk about columns, resins, and membranes. Tarpon Biosystems created the original disposable prepacked axial columns for use at Xcellerex. This technology was subsequently sold to Repligen and Life Technologies. Repligen redesigned the columns and now offers Opus prepacked disposable columns ranging from 50 µL up to 150 L (80 cm diameter). They can be packed with user-specified resin (over 300 resins available) with a range of bed heights from 0.25 to 30 cm. Sepragen Radial Flow columns (Superflo) operate four to 10 times faster at a lower operating pressure with lower pool volumes than conventional axial columns and are available in prepacked formats.

Rigid resins are easier to pack, require much less washing, and maintain bed integrity without wall support. These resins also may be operated at high linear velocities (800 to 1,200 cm/hr) without significant loss of productivity. Pall Danaher is commercializing a fiber-based chromatography platform obtained via the acquisition of Puridify from Cytiva. This fiber format has an open pore structure that allows for convective mass transport, leading to flow rate-independent binding capacities and cycle times of minutes compared to hours for traditional resin chromatography. These membranes and fibers can be used for rapid cycling across a single batch before disposal.

The capture step by affinity resin is the best option for mAbs. Affinity allows for greater reduction of HCP and hcDNA as flow-through. The choice of matrix composition also will contribute to either removal or retention of contaminants and may either eliminate or increase the need for further purification downstream of any step.

The order following the capture step generally is dictated by the elution conditions from the previous step. This reduces manipulation of buffers between steps, avoiding unnecessary TFF steps. Some buffer dilutions may be required to alter the conductivity or pH of the previous elution step, and should be handled in mixing tanks when possible, especially for bind-elute steps. Flow-through steps are usually reserved for the last step prior to TFF since they tend to increase the volume of the drug substance. Use of membranes versus resins in flow -through steps helps to reduce the resulting increase in volume and shorten the TFF concentration step.

Choosing the appropriate molecular weight cutoff for the TFF membrane is critical. For mAbs (150 to 180 kD) it’s usually 100 kD, although there are instances where smaller mAbs (<120 kD) are encountered that require a tighter cutoff (50 to 70 kD). Membranes of 10 to 30 kD simply do not have pore sizes that are large enough to allow most HCPs (<60 kD) and light chains (25 to 50 kD) to pass through.

Lastly, the key driver in any purification strategy is cost. Always consider the manufacturing costs when designing your method. Relatively small increases in cost during PD may result in great savings during manufacturing when chosen wisely. Whenever selecting a particular step (resin or membrane), always consider the GMP manufacturing cost (validation, suite time, buffer consumption, etc.). Always let your critical quality attributes drive the process development. Going the easier or least expensive way early on usually comes back to haunt you later.

References:

BioPlan Associates, Inc. (2019) Top 15 Trends in Biopharmaceutical Manufacturing, in 16th Annual Report and Summary of Biopharmaceutical Manufacturing Capacity and Production, Oct. 2019.

GE Healthcare Life Sciences (2019) Intensified Chromatography Strategies, retrieved from https://www.gelifesciences.com/en/us/solutions/bioprocessing/knowledge-center/intensified-chromatography-strategies.

Lavelle, L (2019) What’s New in Manufacturing: Process Chromatography, BioPharm International, Vol. 32:9.

Pentia, Gonzalez-Perez, Fabiano, Peyser, Vicalvi, Sesay, and Tingley (2016) Reducing Clinical-Phase Manufacturing Costs. BioProcess International 14(11) December.

Peters, RC (2020) Faster, Better Bioprocessing in 2020, BioPharm International, Vol.33:1.

Stevenson, RL (2013) Second-Generation Expanded Bed Adsorption Scales from Lab to Plant, American laboratory 45(2), February 2013.

Vicalvi (2015) “Emphaze™ AEX Hybrid Purifier as the Utility Infielder of Your Downstream Toolbox: A Tale of Two Antibodies”, Guest Speaker, BioProcess International Conference & Exposition.

Vicalvi (2020) “Suppliers' Contributions to Bioprocessing: Review of Who Developed Genuinely Significant New Technologies and Products” in 2020 17th Annual Report and Survey of Biopharmaceutical Manufacturing Capacity and Production, pp.64-79.

About the Author

Jack Vicalvi is a principal scientist and director of downstream process development at J2 Biopharm Associates, a consulting firm, where he advises biopharmaceutical companies on GMP manufacturing processes and provides support for regulatory submissions. Before that, he worked for Pall Life Sciences as a principal scientist and downstream process development manager. Previously, his posts included leadership and research positions at Goodwin Biotechnology, Xcellerex, Exact Labs, Catalent, and Sunovion Pharmaceuticals. His undergraduate study was at St. Anselm College followed by graduate programs in immunology at Southern Illinois University and parasitology at Vanderbilt. Connect with him on LinkedIn.

Get the latest articles from Bioprocess Online delivered to your inbox.