Determining and Optimizing Dynamic Binding Capacity

Bnetto/Stock.Adobe.com

Dynamic binding capacity (DBC) indicates the quantity of a specific biomolecule that can be bound during a chromatography purification step under a specific set of process conditions. Determination and optimization of this value facilitates the development of efficient and productive chromatography operations. While the same general method is used to determine the DBC for many different biomolecules, matrices, and chromatography operations, all of these variables can influence which process parameters are important. Thus, there are many different factors that must be taken into consideration and challenges that must be overcome when optimizing DBCs and ultimately chromatographic performance.

Static vs. dynamic binding capacity

Column binding capacity can be measured in static or dynamic mode. Static measurements performed in a beaker, tube, or 96-well plate provide information on the total amount of protein that a given amount of chromatography media in a given volume of solvent can bind for a given protein concentration. Excess protein is typically used to ensure maximum binding.

DBC is determined under operating conditions (packed column or equivalent membrane/monolith device operated under flow) and is thus much more representative of actual process conditions and directly comparable to large-scale processing. Adding flow decreases the time the molecule has to diffuse into a pore at a given point as it traverses through the column, says Nathan Zeller, scientist II, process development in the Pharma Services business of Thermo Fisher Scientific. That adds a barrier to diffusion that is not present under static binding conditions. As a result, static binding capacity measurements do not characterize frontal binding characteristics that occur along the bed length of a packed column or membrane/monolith device, according to Tom Elich, manager, manufacturing, science, and technology purification process engineering, MilliporeSigma.

“DBC measurements take into account the fact that flow can greatly influence mass sorption (the capture and release of material onto solids/resin),” adds Jacob Brockerman, associate development scientist, AGC Biologics. Static binding capacity measurements also do not take into account that flow rates and other conditions are constantly changing, information that Daming Huang, development scientist, AGC Biologics, says is crucial for optimizing column performance and ensuring that the desired level of protein purity is achieved.

Furthermore, because they are performed under the conditions expected to exist during actual production and under the same specific column controls, DBC measurements provide needed information about the effectiveness of manufacturing controls over a wide range of parameters, including column performance requirements, column volume, and process flow rates or residence time, among others, notes Noah Vartanian, associate development scientist, AGC Biologics. “Designing a purification process based on data from a static binding capacity may lead to rude surprises when the scaled-up column needs manufacturing controls exerted on it, making the purification process perform outside the static binding capacity data,” he comments.

Another benefit of using DBC measurements is the fact they allow for more accurate determination of loading capacities and target protein breakthrough profiles, according to Elzbieta Pawlowska, research scientist, Astrea Bioseparations. “That means chromatography purification steps can be optimized to avoid overloading and subsequent loss of target protein,” she remarks.

Static binding capacity can, observes Tomas Björkman, senior scientist, R&D, Bioprocessing, Cytiva, be used for modeling a dynamic process, but the one measured static capacity result may not match with the capacity obtained in a flow process. “The DBC is lower than the static binding capacity,” he explains, “because of the diffusional limitations associated with chromatography resins. It takes time for the target molecules to penetrate the pores to the binding sites inside the particles.”

Alexandre Boulbrima, program technical lead, Charles River Gene Therapy Solutions, argues, however, that static binding capacity can sometimes still be used to highlight specific chromatography phenomena such as protein displacement. This behavior is characterized by an increase in binding capacity when overloading the column and is driven by the protein/compound of interest having a greater affinity to the resin than other impurities, resulting in displacement of those impurities. “Such behavior can be hard to predict using a DBC characterization approach unless previous data suggest otherwise,” he notes.

The breakthrough approach most common for determining DBC

At its simplest, DBC measurements involve loading a packed column with a known amount of sample at a set flowrate and buffer conditions (i.e., pH, conductivity) that facilitate binding of the target molecule, which, depending on the application, could be the product or an impurity. Load flow-through is collected from the column outlet, and the concentration of the target molecule is measured over the course of loading. When the concentration of the target molecule in the load-flowthrough (C) is the same as that in the feed (Cf), 100% breakthrough has occurred. The data are used to generate a breakthrough curve with load flow-through concentration (Y-axis) plotted as a function of volumetric or mass loading (X-axis). The DBC is commonly defined as the mass loading at 10% breakthrough (C/Cf = 0.1), according to Elich.

The sample loaded on the column may be pure or purified target molecule, notes Björkman. For Zeller, using representative load material is the best approach. “Harvest cell culture fluid (HCCF) is best for evaluating a Protein A affinity resin for the initial capture of a monoclonal antibody (mAb). If evaluating a resin for the subsequent bind-and-elute polishing step, such as cation exchange (CEX), then the capture elution that has been pH- and conductivity-adjusted would serve best,” he explains. Taking this approach will, Zeller says, provide the most accurate results because biomolecular impurities can affect binding of the target molecule.

In fact, Boulbrima contends that the best methods are often limited by their analytical throughput. “Ideally, a fractionation of the load flow-through should be performed, and samples should be analyzed to probe any potential breakthrough of material, indicating the maximum DBC has been reached. Nevertheless, it is important to note that the precision of this DBC value will be highly influenced by the assay precision and overall resolution of the fractionation,” he observes.

The fractionation resolution (the number of fractions to be collected per unit of volume), Boulbrima adds, should ideally not exceed the assay resolution, as it will be difficult to distinguish actual variations between samples versus the assay noise. There is often a tradeoff, however, with the greatest resolution not always being worth the cost, manpower, or time. “Ultimately, the ‘best’ methods are often a compromise of resolution (limited by the assay level of quantification, overall assay noise, etc.) and time,” he concludes.

Zeller agrees that the “best” method to evaluate or determine a DBC will depend on the resources available. “If there is a fast method for accurately analyzing concentration or titer that is readily available, then that would be the best method,” he states.

Impact of residence time for different media types

Regardless of the chromatographic media type—conventional resin beads, monoliths, nanofiber adsorbents, cast membranes, and hydrogels, the general approach for determining DBC remains the same.

The complication for non-resin-based adsorbents, according to Ian Scanlon, subject matter expert, downstream processing, Astrea Bioseparations, is specific to the device which is used to house the material, where the dead volume in the unit is generally much larger compared to a packed bed adsorbent column. “The dead volume must be accounted for in order to accurately measure the capacity of the material for scale-up purposes,” he says.

Dead volume, Scanlon continues, can be determined using a salt transition or non-interactive UV analyte, whereby the measurement of the 50% point of the transition is generally accepted as the dead volume of the housing unit and membrane bed. This value can then be used to determine the breakthrough curve.

Another factor that must be explored when using different media types is the impact of the residence time, which is related to the flowrate. “Binding capacity for resin beads is dependent on diffusive mass transfer, and it is common to characterize binding capacity at several residence times in the range of three to six minutes for bead-based technologies. For membranes and monoliths, which are more dependent on convective mass transfer rather than diffusive mass transfer, the residence time parameter is less critical,” Elich observes.

For many novel adsorbents, notes Scanlon, residence times can actually be significantly reduced. For instance, the residence times for the nanofiber adsorbent developed by Astrea range from 1 to 12 seconds compared to minutes for resin-based technologies. For this reason, Scanlon strongly recommends consulting manufacturer recommendations for flow rates and residence times to use during DBC determinations. Zeller also cautions that the flow rates for monoliths and membranes are typically near the limit of the fraction collection speed, which can create other challenges.

Potential for competitive binding of impurities

The nature of the biologic substances in the feed stream for a chromatographic purifications step will naturally have an impact on the DBC. They can also influence the DBC determination. The biggest concern that must be addressed is the potential for competitive binding effects between the product and impurities present in the feed. If competitive binding occurs, the DBC can be reduced, according to Elich.

“The complexity of feedstocks should be accounted for with all biologicals, but particularly where competitive binding is possible,” adds Scanlon. He notes that the issue is less of a concern with affinity interactions (e.g., when using proteins A, G, and L) because they are highly specific, but for many of the newer targets these types of affinity ligands are not yet commercially available.

Competitive binding occurs when two or more biomolecules have similar physicochemical properties. Elich uses plasmid DNA and RNA as one example. In this case, the molecules have similar sizes and isoelectric points, which creates the risk for reduced plasmid DNA binding capacity when using anion exchange resin. “If the chromatography conditions are not properly optimized, RNA impurity will compete with plasmid DNA for binding sites, limiting available capacity for product binding,” he says.

Ion-exchange chromatography for viral-vector platforms suffer particularly from competitive binding issues, according to Elich. “Many process- and product-related impurities feature acidic isoelectric points that are comparable to the viral vector product, posing risk to competitive binding effects. This makes optimization more challenging as compared to traditional mAbs that feature a basic isoelectric point.

“In these cases, it is important to determine loading volumes with the appropriate feed rather than using a purified target, as this may lead to overestimation of capacity,” Scanlon states. He adds that it is useful to ascertain if the loading conditions can be optimized to avoid binding as many of the present impurities as possible. “A salt concentration and pH screen at the DBC stage can lead to better capacity for the target and cleaner elution pools,” Scanlon contends.

Another important point to remember, Boulbrima emphasizes, is that the product of interest might not always be the determinant of the DBC. “In the case of a polishing step using a flow-through method,” he explains, “the product of interest will flow through by design, whereas impurities will remain bound to the medium. Sizing can then be achieved through targeting the maximum DBC of the impurity(ies) screened for removal.”

Additional considerations needed for new modalities

Finally, Scanlon notes that DBC determination for novel therapeutic modalities for which quantification of the target is not possible using UV alone (e.g., viral vectors, vesicles, etc.) can be problematic. “In these cases, incremental collection of the flowthrough fractions during load and offline measurement of concentrations of the target can be used to determine the breakthrough and calculate the DBC per unit volume,” he says.

Similarly, obtaining DBC values for new mAb-derivatives such as bi/tri-specifics is not as straightforward as it is for conventional mAbs and recombinant proteins, according to Zeller. “The mismatched mAb impurities (homodimers, half mAbs, mismatched light chains, etc.) tend to exhibit similar binding characteristics to the target molecule. It is therefore harder to get a DBC value that maximizes the binding of the molecule while reducing these specific impurities.

As importantly, Scanlon notes that stability of novel therapeutics is often an issue, and thus collected flowthrough fractions should be treated appropriately to avoid inaccurate quantitation.

Selection of monitoring methods important

The methods used to monitor breakthrough must be appropriate for the type of chromatography being performed. “The principle for determination of DBC is independent of the target molecule, but the analytical methods required for detection of breakthrough may differ,” notes Björkman.

For instance, for viral-vector purification, there is a debate about which assay to use for DBC determination and optimization. “In the case of lentiviral particles, both total particle titer (often achieved via an ELISA [enzyme-linked immunosorbent assay] against the p24 capsid protein) and infectious particle titer (cell-based assay) could be used, but both have displayed great levels of inconsistency so far,” Boulbrima comments.

Methods typically differ for capture vs. polishing steps as well. “HCCFs are full of impurities, so a titer method is typically required to get the concentration values for breakthrough analysis for capture steps. For cleaner material loaded onto polishing resins, however, a microplate spectrophotometer can often be used to get the concentration values of the effluent fractions,” Zeller comments.

Elich agrees.“Because of the low feed purity for capture steps, the UV signal at the column outlet can become saturated with unbound impurities (i.e., host-cell proteins [HCPs], DNA, and cell debris), making UV absorption an imprecise measurement tool for this application. A better method might be to collect fractions for concentration measurement by offline high-performance liquid chromatography (HPLC) analysis.” For post-capture polishing steps, however, online UV absorption might be a viable analytical tool for breakthrough curve assessment.

Here again, it is important to remember that for polishing steps in flow-through mode, the goal is to understand impurity binding capacity as opposed to product binding capacity. “Molecule-specific analytical tools should be used to construct breakthrough curves of the target impurities that bind the chromatographic matrix as the product flows through (i.e., HCPs, DNA, virus, product-related impurities, etc.),” Elich says. Typically, ELISA or HPLC methods are used for offline analysis.

Competitive binding not the only challenge

While competitive binding is generally considered the biggest challenge to optimizing DBCs, there are other issues that must be addressed. For instance, Björkman notes that the binding capacities of affinity steps are usually more robust within a certain pH and salt range and are therefore not necessary to optimize in the same way as non-affinity steps, which require careful selection of pH, salt concentration, and salt type. Boulbrima adds that capture steps are also usually robust with respect to the compositions of the feed streams employed, while the performance of ion-exchange and hydrophobic interaction chromatography steps can often be far more influenced by the feed stream quality and composition.

Elich also points out that it is important to consider the impact of media chemistry on DBC determinations. Binding capacity experiments may be designed differently for affinity, ion exchange, and hydrophobic interaction technologies, for instance. As an example, he notes that the DBC for affinity chromatography is less dependent on conductivity, whereas that for ion exchange and hydrophobic chemistries is heavily dependent on this feed property.

Using dilute crude samples is also more difficult than using purified material, according to Björkman. “Often when using a crude sample, you need to collect fractions and perform off-line analysis and that adds to the work needed and also to the uncertainty,” he explains. Björkman also observes that limited access to sample material may require use of miniaturized techniques, small columns, or even batch experiments to estimate binding capacities.

In addition to the flowrate, Gráinne Dunlevy, R&D director delivery, Astrea Bioseparations, points to the importance of considering the delta column pressure, which increases with flow rate for traditional adsorbents. This issue is not a concern for next-generation matrices such as nanofibers because there is little back pressure, which is one of the reasons higher flow rates are possible.

Another challenge noted by Zeller relates, once again, to whether the operation will be run in bind-and-elute or flowthrough mode. Bind-and-elute is generally easier, because monitoring involved detection of the single target molecule. Flowthrough mode can be trickier, he observes, because of the heterogenous pool of impurities, which typically requires a cell line-specific protein detection assay, such as an ELISA for host cell proteins. If the flowthrough operation will be claimed as a viral clearance step, then it is also necessary to determine how much material can be loaded without virus breakthrough.

Optimizing DBCs

Optimizing DBCs can be achieved using a variety of approaches. In some cases, static binding experiments performed in 96-well plates are useful as a first step because they enable high throughput screening of relative binding capacities for a wide range of traditional resins and/or buffer conditions, according to Elich. He also notes that more recently predictive modeling software has emerged for optimizing the performance of a chromatography step using mechanistic and thermodynamic models.

Screening is helpful because optimization of the DBC requires identification of parameters that affect it and then optimizing those parameters, according to Brockerman. “The specific set of important parameters varies from process to process (resin vs membrane, affinity vs ion exchange, etc.), but typically include load buffer composition (pH, salt, etc.), load concentration, flow rate, and residence time. Given the number of parameters, Brockerman notes that multivariate analysis (design of experiment [DoE]) approaches have proven exceptional for efficiently identifying relevant parameters and their optimization.

Ben Galarza, research assistant, Astrea Bioseparation, adds that the DoE approach is applicable across all media and purification steps and is a powerful tool because it allows selection of both numeric and categoric factors. “For instance,” he observes, “different types of media can be input as categoric factors alongside any numeric factors if there is uncertainty as to which media to select.” He does note that DoE strategies may be more relevant for the development of chromatography processes for new modalities.

In addition to optimization of typical process conditions, other strategies highlighted by Huang include column packing and particle size optimization, use of the appropriate column conditioning steps, such as equilibration, and column scaling to the appropriate size. “These strategies are applicable across different media, purification steps, and drug substance types, although the optimal approach may vary depending on the specific needs of each purification process,” he says.

More general approaches to optimizing the dynamic binding capacity of chromatographic purification steps in biopharmaceutical manufacturing noted by Huang include the use of multimodal chromatography, which can increase the DBC while reducing the number of chromatography steps; membrane chromatography, which can significantly increase the DBC while reducing the amount of resin required for purification; continuous chromatography (vide infra); and automated purification systems, which can increase the DBC while reducing risk of operator error and increasing process consistency and reproducibility.

Zeller agrees that using a liquid handling robot capable of performing chromatography is the most efficient method for optimizing the loading conditions that then effect the DBC, especially during complete development of a polishing step. “This approach can screen multiple resin types, buffer conditions, and molecules in one programmed method and can usually be complete in less than 24 hours compared to
days or weeks using a conventional fast-protein liquid chromatography instrument,” he says. Zeller does note that these systems are not yet compatible with membrane chromatography.

Achieving consistent DBCs during scaling

The scaling of chromatographic purification steps can impact the DBC in many ways. “As processes are scaled up, certain physical parameters (such as decreased capillary action at large scale) can start having a much greater influence on the chemical interactions in the exchange between static and mobile phases,” Boulbrima explains. This behavior, he says, can potentially be predicted for a specific biomolecule and a specific chromatographic support, using either a robust scale-up characterization strategy or a robust small-scale model qualification strategy and/or historical data at all scales. “The latter will especially be key for continuous improvement of the platform, knowledge gathering, and outlier characterization,” he contends.

Maintaining an optimum DBC when scaling chromatography steps is generally possible if the bed height and residence time are kept constant by increasing the column diameter, according to Björkman. “While there are some cases where the residence time is of no importance within a reasonable range, maintaining the same residence time used for determination of DBC as processes are scaled is always a good rule of thumb,” he says.

A properly designed scale-down tool should enable consistent binding capacity between small- and large-scale operations, agrees Elich. In practice, he notes that it is common to employ a safety factor to account for any process variations between scales, resin lots, cleaning cycles, or product batches. As an example, he notes that during operation, a chromatography matrix may be loaded only to 80% of the DBC measured at 10% breakthrough (80% of DBC10%).

Dunlevy agrees that because the DBC for most chromatographic processes depends on the residence time, keeping it constant during scale up is important. She suggests, however, that it is possible to keep the bed height and linear flow rate constant or adjust for any increase in the bed height by increasing the flow rate. The downside of these approaches, Dunlevy notes, is that increasing the flowrate and bed height lead to higher delta column pressures, and thus changes are limited by the properties of the resin being used. Newer, alternative chromatography matrices such as fiber have an advantage here, because they can be kept at very short bed heights with low delta column pressures and thus can be scaled more easily.

Continuous chromatography

Running chromatography steps in continuous mode does not have an impact on DBC determination and optimization. “Continuous chromatography with columns operated in overload mode (i.e., a second column collecting the target flowing through the first column) will not complicate the optimization,” remarks Björkman. “Conditions are optimized using the first column and then depending on system, the operation can be quite simple, especially if on-line sensors such as UV-cells are placed before and after the column and software controls the flow path,” he adds.

Loading multiple columns in series allows for not only continuous loading, but a DBC that is closer to the media’s static capacity and without sacrificing yield, according to Elich. In fact, it enables use of the full DBC of the column, remarks Dunlevy. “In standard batch chromatography, loading is typically to 80% of the DBC to minimize any breakthrough during the loading phase. That decreases the yield and prevents leveraging of the full DBC of the column,” she says. Elich notes that typically only 50–60% of the static binding capacity is used in batch chromatography. In continuous chromatography, because the breakthrough is loaded onto the next column in series, reaching nearly 100% of the DBC of the column is possible, Dunlevy adds. In the end, continuous chromatography actually lessens the importance of the DBC due to the efficiency of the processing, observes Zeller.

In fact, Zeller emphasizes, one of the major drivers for increasing binding capacity is to reduce the number of cycles needed to purify a product. “Since the simulated moving bed process is quick and efficient, multiple cycles can be completed in the time it takes for one conventional batch chromatography unit. Therefore, DBC is less important in that aspect,” he comments.

For large-volume processes and/or when cycle count is important, however, Zeller stresses that high DBC values help to concentrate the product and reduce the number of cycles that would otherwise be needed, which results in a lower buffer footprint.

There are some caveats. Design of continuous chromatography operations requires accurate characterization of the complete breakthrough curve profile during DBC assessment, which according to Elich is best accomplished using a column bed height representative of process-scale operations (typically near 10 cm). The life cycle of the chromatographic support (resin, monolith, others) must also be characterized and the right cleaning procedure put in place, Boulbrima states. These steps can be achieved via a resin-reuse study, which would not differ much from what is required for conventional processes using chromatography multi-cycling.

A few tips about DBCs

Developers of chromatography purification processes for biologic drugs should think about a few other aspects of development activities to ensure greater likelihood of success. For instance, the DBC for any specific biologic should be determined at the start of a process and characterized over as many of the variables present in the process as possible, according to Scanlon. In addition, if any part of the process is changed at any time, reevaluation of the DBC should be performed to identify any impact of the change. Furthermore, the DBC should be checked during lifetime studies that reflect the use-case of the adsorbent.

Specifically for polishing (IEX, HIC) steps, Zeller recommends that the load material be adjusted to the pH and conductivity conditions of the first wash buffer to mitigate product loss with buffer matrix transitions.

Finally, Elich notes that DBC is only one factor considered when optimizing bioprocess chromatography steps. Purity and yield are two other important factors, and it is possible that conditions that provide an optimal DBC do not optimize purity and/or yield. “A balance of these parameters must be achieved,” he states.

Björkman agrees that the DBC is an indication of the maximum quantity of the target molecule (or impurity/impurities) that can be bound under certain conditions (buffer composition and column residence time), while the optimal load value also takes the desired purification objective into account. “If the chromatography step is used to remove impurities that also bind to the column, it may be necessary to reduce the loading significantly below the DBC of the pure target molecule. Consequently, the column load is often one of the optimized factors for a purification step,” he concludes.

About the author

Cynthia A. Challener, PhD is a contributing editor to BioPharm International.

Article details

BioPharm International
Vol. 36, Number 4
April 2023
Pages: 10-14, 34

Citation

When referring to this article, please cite it as Challener, C.A. Determining and Optimizing Dynamic Binding Capacity. BioPharm International 2023 36 (4) 2023.