• Elimination of immunogenicity to the Fc region
• Easier penetration into tissues due to their smaller size, improved target access and bio-distribution
• Reduced cost to produce via microbial and yeast systems and elimination of viral clearance requirements

Traditionally, manufacturers have leveraged Protein A as the principal method for mAb purification.
Antibody fragments, however, do not have a similarly defined, well-established template for purification. Our experts weigh in on the latest developments for the purification of antibody fragments.

Lucentis® is expressed in Escherichia coli cells, a microbial system, as opposed to mammalian cell culture that is typically used in production of full-length monoclonal antibodies (MAbs). As such, the task of developing a purification process is presented with several unique challenges. The Fab fragments are typically secreted into, and sequestered in, the periplasmic space of the microbe. Therefore, some type of chemical or physical cell disruption step is needed, followed by a clarification step that is robust enough to handle the high solids that are likely in a bacterial fermentation process.

With full length monoclonal antibodies, there is generally a consensus that Protein A chromatography media is an excellent choice for the capture step and can be broadly applied to multiple MAbs in a platform process approach. For antibody fragments like Lucentis®, Protein A-based media isn’t an option so Fab fragments may be more difficult to develop as platform processes. Protein L, a bacterial protein which has a high affinity for the kappa light chains present in many Fabs, was an option but it was very expensive at the time and not practical for implementing in a large-scale commercial process. Over the years, I’ve worked with multiple vendors that have tried, unsuccessfully to date, to develop a selective, high-affinity ligand capable of binding light chain. The hope was to develop a low molecular weight ligand with the benefits of Protein A-based , media, without the use of protein-based ligands. Although unsuccessful to date, I expect the technology will be available sometime soon. The ideal resin would be cost- effective, with reasonable binding capacity, and highly selective for Fab fragments; media that would serve the same role as Protein A does for full-length antibodies. As an alternative to affinity-based approaches, cation-exchange chromatography may work as the capture step, but these steps would likely have to be fined-tuned each time for each Fab. This type of effort will clearly be helped by recent advances in high-throughput process development so that the understanding of acceptable operating space, which may have taken months previously, may now be done in weeks.

In addition to the difficulty in establishing a platform capture step, other challenges can exist when dealing with Fab fragments. A common difficulty is that light chains are frequently over-expressed during execution of the upstream process. Overexpression of light chain typically leads to the formation of light chain dimers, a product variant that can be difficult to fractionate as its chromatographic behavior is typically highly similar to the Fab itself. With a commonly-employed capture method, such as cation-exchange chromatography, you may have strong competition from the dimers for capacity on the column. In these instances, a flow-through step capable of removing the light chain dimer prior to the first bind and elute step could be very valuable.

Complicating the purification is a number of factors, one of which is the possibility to use microbial expression systems to express the target molecule. It is attractive while the need for virus clearance is generally not present and the productivity is very high. But, with a microbial system, there are toxins that must be dealt with and also the target molecule unfolding/refolding steps – a process that can have a significant impact on yield.

With the diversity of the fragments, there isn’t a templated approach but there can be tendencies which are explored. The first part of the purification platform – opening the cells, diafiltraton or centriguation and solubilization — will remain very much specific to the target molecule and optimized for the expressed fragment. But the capture, polishing steps could be somewhat templated if some more universal target molecule domain remains enabling the use of the protein A or G affinity chromatography. An important consideration must be the refolding buffers that are used as some of them contain substances that might not be compatible with affinity chromatography.

Alternatives to protein A or G can certainly be ion exchange, hydrophobic interaction, and hydroxyapatite chromatography. Once those decisions have been made and the fragment is successfully captured and concentrated, the rest of the process shouldn’t be much of the challenge.

Also, due to the smaller size of fragments, it is typically better to use small beads to ensure better selectivity.

Clearly the main difference when purifying Fab fragments as compared to full length antibodies is that protein A cannot always be used as the capture step as it has low capacity. High Throughput Screening must be done to identify the optimal approach for capture and that could be mixed mode, cation or anion exchange as well as hydrophobic interaction chromatography. We currently conduct screening in 96-well plates with a goal of identifying the simplest process, which also tends to be least expensive. The polishing step can then be similar to that used for a full-length monoclonal.

When working with antibody fragments, what can be easier is the lower level of contaminants in your starting material as compared with monoclonals derived from CHO cell culture, for example. With fragments produced in bacterial systems, the ratio of contaminants to your molecule will be lower due to higher productivity compared to monoclonal antibody.

With resin bead chromatography however, the solute must diffuse into the pores within the beads in order to reach available binding sites.  Membrane chromatography technology evolved from the need to overcome this limitation.

Membranes tend to be much more efficient than beads with faster separation times, but with a tradeoff in capacity measured as the amount of material that can be absorbed per liter of device.

Our experts explore the pros and cons of each approach.

Applications where the target protein binds and elutes from the membrane are less common but do exist. Membranes operated in the bind and eluate mode may offer increased benefits for applications where 1) the target protein concentration is low and volumetric loading is high, 2) continuous processing such as simulated moving bed is being used (decreased loading zone time), 3) quick processing is required due to poor protein stability, 4) there are IP issues with using tradition chromatography in a given application, such as with biosimilars, or 5) low cost purification is required for example in early phase trials.

One challenge of using membranes is the relatively small device size. Chromatography columns can easily exceed 100L of resin whereas membranes are far more limited in size. This small device volume results in the need to cycle multiple times. Although this doesn’t necessarily increase cycle time due to the high flowrates it does introduce process questions that need to be addressed. These include, but are not limited to, cycle to cycle carryover and cleaning, membrane life time, product pooling strategies and product sampling frequency.  That said, as long as you can achieve capacities that are reasonable, membranes can support many applications.

In summary, single-use membranes offer more control over the cost of goods and reduce operational effort including, cleaning validation, column packing, buffer prep, etc.  This can have a positive impact on cost of goods for early phase processes although the opposite maybe true for high volume commercial processes.

All else being equal, ideal matrix selection (bead-based resins vs. membranes) is highly application dependent. Due to their porous structure, bead-based resins are typically characterized as having high internal surface area, resulting in high binding capacities for molecules able to readily diffuse into the beads. Consequently, bead-based resins are best suited for capacity-limited applications, such as the traditional bind-elute purification of molecules able to readily access the internal surface area of the beads.

In contrast, membrane-based media are usually characterized as having significantly lower surface area (binding capacity) compared to bead-based resins, although this surface area is readily accessible with minimal diffusional resistance – even for extremely large molecules. As a result, membrane-based media are ideally suited for flow-through polishing applications where low concentration impurities must be efficiently removed from the target product. Since this application is not capacity-limited, small membrane devices can be utilized to effect the separation.

Membrane-based media are also well suited for bind-elute applications for extremely large molecules (plasmids, viral vaccines, etc.).  These large molecules are unable to access the internal structure of tradition bead-based resins, resulting in extremely low binding capacities. For membrane-based media, the binding ligands are readily accessible within the convective flow field. Consequently, the binding capacity for these large molecules can often be 10x higher on membrane-based media than on bead-based resins.

In an effort to address this, platform technologies, in which prior knowledge was collected and leveraged for new molecules, were identified. Platform molecules can then proceed through development using standardized, manufacturing processes. In contrast , a customized approach is created for non-platform molecules, for which more techniques and conditions have to be evaluated.

Our experts weigh in on how they are applying platform technologies, best practices, and the benefits they have experienced.

We have standardized on a number of process elements including affinity chromatography such as Ni-IMAC and analysis including SDS-PAGE and tryptic digestion of SDS bands, and mass spectrometry analysis for protein identification of product and impurities. Size exclusion chromatography is applied often as final purification step, if required.

We have published on screening for parameters for optimal protein purification by liquid chromatography and application of parameters determined by screening experiments for sample displacement batch chromatography.

The chromatographic steps following affinity chromatography are customized each time. Customization is achieved by rationale parameter screening.

Our use of platform technologies has evolved over the years. At the beginning of the platform development, the screening platform was established, followed by the integration of fast read out system for monitoring the screening results. Sample displacement batch chromatography based on rationale screening is the latest platform which we developed.

Through use of platform technologies we have seen that the purification process is much faster in comparison to former approaches. And there are no disadvantages because the screening platform provides a large degree of customization.

Most companies we work with have a defined toolbox of resins for the three step process and select from that predefined toolbox with each new molecule. This limits the parameter space to a certain extent. By defining a certain sequence of preset technologies, one can experience more rapid process development and ultimately increase speed to market.

The platform approach works well for monoclonal antibodies as they all have a common parameter – they all bind to protein A during the capture step. Use of templates outside the monoclonal antibody space still needs to be explored. With microbially-expressed proteins or biosimilars, each is different and this limits the ability to establish a set of platform technologies. Use of expression tags such as histidine can enable development of a platform as some type of common, generic feature is needed.

Even with a platform in place, it is important to evolve it as necessary to incorporate new technologies and add new parameters to the toolbox.