Platform Technologies

Platform technology is becoming a popular industry approach for bioprocessing, but just how are companies using it? Pharmaceutical Technology talked to industry experts to gain insight: Morrey Atkinson, PhD, CSO and vice-president of R&D and Drug-Substance Manufacturing at Cook Pharmica; Peter Moesta, PhD, senior vice-president of Biologics Manufacturing and Process Development at Bristol-Myers Squibb; and Jim Powell, business development manager at Ashai Kasei Bioprocess.

(IMAGE: GREGOR SCHUSTER/PHOTOGRAPHER'S CHOICE RF/GETTY IMAGES)

PharmTech: How might platform technologies be applied to upstream and downstream processes? Is one easier than the other?

Atkinson (Cook): It is not easier to develop platforms for either upstream or downstream, it is just different. The main difference in developing platform processes for either is that, in most cases, one develops upstream processes for the cell line and the expression system, while downstream processes are tailored to the molecule itself. If the molecules are of a similar type, then the downstream process becomes easy to develop.

In terms of difficulty, the cell lines and expression systems are inherently variable, and clone-to-clone variability adds to the complexity. Scale factors are also more difficult to control in cell culture and fermentation. In general, upstream therefore probably poses a slightly greater challenge, assuming that the molecules are in a given class or category.

Moesta (Bristol-Myers): With today's level of know-how in molecular biology and expression, platform technologies are easier to develop for upstream processes. Identification of a preferred strain or cell line for microbial or mammalian expression, combined with a well-developed expression vector, is the first step in establishing a production platform. This step allows for the use of standardized fermentation or cell-culture conditions requiring limited media and feed optimization. The use of platform expression systems and upstream conditions allow for the generation of significant process experience and forms the basis for developing downstream platforms to the extent possible.

It is easiest to develop a standardized process for initial downstream steps (e.g., centrifugation and depth filtration for cell-culture products). For monoclonal antibodies (mAbs), where the Fc protein domain–Protein A interaction can be exploited to capture the protein from clarified cell-culture broths, additional platform steps are possible (e.g., Protein A-based affinity chromatography and viral inactivation and filtration steps). The final purification steps (i.e., polishing) need to be tailored to the particular antibody at hand and usually require individual optimization. For other proteins, downstream processing becomes less amenable to the platform approach. Individual process steps can be standardized, but will need to be pieced together and optimized on a case-by-case basis.

BMS is developing molecules to which we apply platform-based approaches, including antibodies and adnectins. But even when dealing with well-defined classes of proteins, key challenges for establishing production platforms result from unique properties of individual proteins, such as charge heterogeneity, differences caused by post-translational modifications, and stability. These unique properties can impact both the cell's ability to express a correctly folded and stable protein as well as purification of a homogeneous drug substance.

PharmTech: Could a platform for purification accommodate variations between mAbs? Is it possible to develop a purification platform for various classes of products (e.g., mAbs and enzyme products)?

Atkinson (Cook): Platform purification processes must deal with both process- and product-related impurities. With antibody processes, the process-related impurities tend to dominate the development of the platform. Removal of host cell proteins (HCP), in particular, is usually a primary driver.

For the product-related impurities, most antibody processes are usually dominated by the removal of higher-molecular weight aggregates, followed by clipped forms and other charge variants. This is why so many platforms use an affinity step, followed by some combination of ion-exchange and/or mixed-mode separation.

It is important to note that a platform process for purifying antibodies must accommodate both charge and hydrophobicity variation between the molecules themselves. The constant regions of most immunoglobulins is consistent in physical and chemical behavior, but single amino acid changes in variable regions can drastically shift either the isoelectric point (pI) of the protein or the relative regional hydrophobicity. So the process must be able to remove a wide range of charge variants as well as various hydrophobic species (e.g., aggregates.)

Moesta (Bristol-Myers): mAbs, Fc-fusion proteins and adnectins developed by Bristol-Myers Squibb have large conserved regions, resulting in physical properties that allow one to achieve the vast majority of purification using platform technology. Charge heterogeneity in the variable region and post-translational modifications then require individually optimized polishing steps. For antibodies, this is usually a combination of an anion exchange step (either in flow-through or bind and elute mode) coupled with a second chromatography step. One must evaluate the remaining purification objective, select the best method, and optimize it. Viral inactivation and filtration steps, as well as diafiltration and concentration steps can be standardized and made to fit with a drug-product formulation platform if available.

PharmTech: Is it possible to develop a purification platform for various classes of products (e.g., mAbs and enzyme products)?

Atkinson (Cook): The challenge in developing platform processes that cover various classes grows as molecular diversity grows. Antibody and antibody-like fusion proteins can be treated as a broad class, but enzymes and other recombinant proteins will have very different molecular characteristics. So outside of broad platform generalities, such as no more than 3-4 columns, all aqueous processing with standard buffers and salts, standard viral filtration systems, and so forth, the platforms will otherwise most likely be quite divergent for different classes of proteins.

PharmTech: Can purification platforms accommodate the rising titers that upstream processes are yielding?

Atkinson (Cook): The rising titers are both a blessing and a curse for downstream unit productivity. The capacity of most chromatography resins is basically sufficient for the increased titers, but the buffer consumption and the throughput become a challenge with very high titers. In this case, limiting the number of unit operations, for example, moving from a three-column antibody process to a two-column process becomes much more attractive. Engineering solutions, such as buffer blending and even possibly simulated-moving bed chromatography can also be considered to manage the increased productivity.

Most downstream unit operations, with the exception of viral filtration, are inherently scalable to an industrial scale. I believe that the biggest downstream bottleneck for high-titer processes has become viral filtration. Most chromatography unit operations can scale effectively, but the need to use an expensive, low-throughput filter can create an inefficient bottleneck in the overall purification process for mammalian cell-derived products.

Scale-up challenges include the high upfront cost for consumables (e.g. resins, bags, filters) as well as the challenges with liquid handling. Bulk volumes of liquid and intermediate holds are inherently inefficient. Engineering solutions for liquid transfer, mixing, and minimizing storage of liquids should be explored.

Moesta (Bristol-Myers): The technology available today can accomplish the manufacture of proteins up to the metric-ton scale. However, few products to date require this large scale of manufacture.

Increasing demand for proteins, combined with higher titers in fermentation, can enable implementation of alternative technologies, such as protein precipitation and crystallization. These technologies provide a means to improve purification throughput while significantly reducing cost. Some examples include blood fractionation products and recombinant insulin.

A key scale-up challenge is chromatography because there are physical limits based on resin-flow characteristics (e.g., back pressure and compression). The next step for industry is the use of simulated moving-bed technology, which can increase throughput.

Powell (Asahi): Process engineers can accommodate rising titers using a combination of liquid handling systems and modern virus-removal filters. Because the ability to rapidly and reproducibly create accurate buffers in a minimal footprint is a common bottleneck during downstream processing, Asahi Kasei Bioprocess offers IBD inline buffer dilution systems to generate on-demand diluted buffers for capture, polishing, and virus-removal. Customized skids are easily integrated with existing equipment to improve purification efficiency.

Additionally, next-generation virus-removal filters facilitate reliable processing at concentrations of up to 50 g/L. Before processing high titers, basic physics of production must be considered. Such factors include the viscosity of the feed material, mechanism of mass transfer, and filter efficiency. Purification becomes easier as the ratio of contaminants to product decreases yet caution must be used as high product levels often reduce cell viability.

PharmTech: With regard to scale up, how do downstream process platforms perform? Are there limitations?

Moesta (Bristol-Myers): Large plants, such as Bristol-Myers' Devens plant in Massachusetts, can have long piping runs between pieces of equipment with significant hold-up volumes. If appropriately designed, process piping either drains by gravity or can be blown out with compressed air, minimizing losses. Filter housings often require water flushes for adequate yield recovery. If properly optimized, large-scale process performance can meet or exceed that observed at smaller scale.

Powell (Asahi): Scaling out, as opposed to up, is the preferred approach. During scale up, a facility transitions to larger diameter columns and filter housings before launching trains of production units in parallel. Besides reagent disposal, additional challenges include space as well as the use of water and buffer.

Holding tanks may be required for byproducts that cannot be released directly into the environment. However, properly designed chromatography systems from Asahi Kasei Bioprocess can reduce the ratio between the hold-up volume and the filter or liquid chromatography (LC) column volume to minimize the waste burden and improve operational efficiency.

PharmTech: Do these scale-up problems require customized solutions?

Powell (Asahi): Obstacles created by process scale up require customizes solutions to a certain extent, especially with respect to automation. When a company moves forward with commercial production, a plant-wide distributed control systems have historically been the preferred method to control and gather data from each step in the process. But for smaller scale production. such as orphan drugs, "islands of automation" are still preferable.

Finer, more accurate monitoring of this nature streamlines operations and enables tanks to open to skids at the proper time. Distributed control systems provide greater access to information in a manufacturing plant, thereby allowing euipment-related problems to be identified and addressed prior to impacting production.

PharmTech: Looking ahead, how can platform technologies for analytical methods be improved? Can methods for detecting contaminating proteins, host-cell proteins, and protein level be standardized? What new technologies or methods could help?

Atkinson (Cook): Analytical methods, by definition, should theoretically be amenable to platform standardization. As mentioned, the primary purpose is to detect process- and product-related impurities. The challenge for process-related impurities is that each upstream platform produces different impurities, such as type and amount of HCPs. Unfortunately 'generic' commercial kits are often poor substitutes for process-specific detection methods, but do serve a purpose when used consistently in a platform.

For product-related impurities, the challenge is similar to that for downstream processing, and depends on specific molecular variants. In addition, the analytical technologies employed are not yet standardized. Charge variants, for instance, can be detected by at least four different methods, none of which effectively discriminate amongst several types of variants (e.g., sialic acid content, deamidation).

If the platform methods are developed in parallel with the process, and used and controlled consistently, then they can be useful within the portfolio they are employed. Process-related impurities are better understood and controlled, and minor modifications can be made to address product-related impurities. However, the relative utility of the platform is lessened when applied more broadly across product portfolios.

Ideal analytical methods would both separate and identify unique molecular species. A high-throughput, quality-control friendly functional equivalent to an LC–MS method would be desirable.

Moesta (Bristol-Myers): Most straightforward analytical methods such as A280, capillary electrophoresis, polyacrylamide gel electrophoresis, isoelectric focusing, and size-exclusion high-performance liquid chromatography, are flexible and lend themselves for upstream and downstream analyses. More complex methodologies, particularly for unique post-translational modifications and potency, are not as easily standardized, particularly those requiring high-end analytical endpoints such as mass spectrometry (MS), nuclear magnetic resonance (NMR), surface plasmon resonance for binding kinetics, and cell-based bioassays.

Once projects progress to the clinical-trial stage, it is advisable to take a closer look at standardized methods and optimize them for the molecule at hand.

The first requirement for being able to facilitate the use of an analytical platform is that the master cell line, expression vector, and upstream and downstream process steps are standardized as much as possible. The better characterized and standardized the process, a combination of anti-sera reactive against known impurities and HCPs can be created from premade anticontaminant libraries to provide sufficient coverage and sensitivity. Protein-A detection methods are relatively easy to platform while HCP methods tend to be the most challenging.

Newer surface-plasmon resonance instrumentation is providing for significant improvement in the throughput and robustness required for ligand binding and binding kinetics assays. With particular molecule classes (e.g., mAbs), standardization of common reagents and capture approaches can improve and simplify the method development of specific binding activity method platforms.

For cell-based bioassays (potency), the use of common cell-based systems, either off the shelf or specifically designed, and activity read-outs for classes of activities (e.g. cytokine production, cell migration, etc.), can significantly reduce the amount of de novo method development. Also, the standardization of read-outs such as chemiluminescence or enzyme-generated colorimetric measurements in a microtiter plate format can further improve throughput.