Novel platforms such as antibody derivatives, peptide based therapies, gene and stem cell based therapies are gaining foothold in the market for several reasons, including the need for better Pharmacokinetics (PK)/ Pharmacodynamics (PD), improved potency against disease targets, ability to treat more than one aspect of a disease simultaneously, better and cheaper production processes, reduced side effects and the biosimilar cliff.

In this article, we will focus on 3 types of antibody derivatives- namely Bispecific Antibodies (BsAbs), antibody fragments (Fabs), and fusion proteins. We will include an overview of each and discuss the typical downstream processes, highlighting specific process challenges. Scale up considerations will also be included.

1.0 Introduction
Monoclonal antibodies (MAbs) continue to dominate in terms of the class of therapeutics for the biotechnology industry. However, the overall trend in the biotherapeutics includes transitioning towards molecules that have higher value and improved bioavailability. Traditional Mabs are altered to achieve this goal, and antibody variants such as antibody fragments (Fabs), bispecific monoclonal antibodies (BsAbs), and fusion proteins are being explored.


The term Fab Antibody fragment is self explanatory- it is the Fab fragment from the variable region of an antibody [1]. A bispecific monoclonal antibody (BsAb) is composed of fragments of two different monoclonal antibodies that bind to two different types of antigens [2]. Fusion proteins are produced from gene fusion techniques that allow the production of recombinant proteins featuring the combined characteristics of the parental products [3].

In terms of common expression platforms, Fabs can be expressed in both mammalian and bacterial expression systems. Bacterial expression system is more common for Fabs and they can be expressed in E.Coli as either inclusion bodies or soluble (soluble being more common). BsAbs and fusion proteins are more typically expressed in mammalian cell cultures.

A typical downstream process for these antibody variants consists of

Fig 1: Typical Downstream Process (click on image to enlarge). Bioreactor/Fermentor > Harvest/Lysis (if bacterial)/Clarification > Capture > Polishing > Virus Clearance (if mammalian cell line) > UF/DF > Final Sterile.

In this article, we outline some of the unique requirements and challenges posed by these antibody variants in terms of recovery, purification, and scale-up/process transfer.

2.0 Part 1 – Recovery

Fig 2: Recovery: Bioreactor/ Fermentor > Harvest / Lysis (if bacterial) / Clarification (click on image to enlarge)

The vast majority of the current therapeutic antibodies including BsAbs and fusion proteins are still produced in mammalian cell lines in order to reduce the risk of immunogenicity due to non-human glycosylation patterns [1]. However, Fabs are more commonly produced in bacterial (E. Coli) due to their smaller size and economic considerations. Bispecific antibodies without any glycosylation could be successfully produced in bacteria as well.

For mammalian cell cultures (used for BsAbs and fusion proteins), normal flow depth filtration can be used for primary and secondary clarification steps for process volumes ≤ 2,000L. Depth filtration has also been shown to assist with removal of impurities such as HCP and DNA and improve downstream filter and column capacities. As titers and cell densities increase, the use of agents such as flocculation polymers/ acid precipitation are becoming more common at harvest.

For bacterial expression systems, clarification is often one of the most challenging steps. For soluble proteins, microfiltration- tangential flow filtration (MF-TFF) is often used instead of normal flow filtration. However, centrifugation followed by normal flow filtration (NFF) can be evaluated. Typically the TFF yields higher product recovery and is more economical. There is also a re-newed interest in older technologies such as using DE as a body feed for clarification. With secreted proteins, whole cells are separated from the fermentation broth and the particle size is thus larger; as a result, microfiltration (using TFF), centrifugation and normal flow filtration are all viable options. Endonuclease agents can also be used prior to clarification to digest DNA and RNA and to aid in the efficiency of the clarification process [2].

For sterile filtration post clarification, the capacity is influenced by the bioreactor/fermentor media components. Symmetric PVDF membranes are better suited for the sterile filtration of PEG and hydrolysate-containing media types. Asymmetric PES membranes are also available and can be evaluated. Sterile filtration step should be optimized with respect to product recovery, capacity and operating flux.

3.0 Part 2 – Purification

Fig 3. Purification: Capture > Polishing > Virus Clearance (if mammalian cell line) > UF/DF > Final Sterile (click on image to enlarge)

Protein A, followed by cation exchange and anion exchange, can be successfully used for the purification of BsAbs, fusion proteins, or Fabs containing the Fc region. For molecules that do not contain an Fc region, capture is typically achieved using cation exchangers or mixed mode resins in bind/elute mode depending on the molecular characteristics of the target protein. A subsequent polishing step for improving the resolution generally follows the capture step. This polishing step could be ion exchange (IEX) or hydrophobic interaction chromatography (HIC) depending on the previous step. And sometimes a third chromatographic step is required, depending on the separation results from the previous steps. In addition to resins, membrane adsorbers are also used for the polishing steps in (usually) a flow-through mode.

Virus filtration is not needed for bacterial expression systems due to the absence of adventitious viruses. For the proteins expressed in mammalian cell cultures, demonstrating viral clearance is a regulatory requirement. Some fusion proteins or BsAbs can be similar in size to virus filter membrane pore sizes (20nm), leading to significant process challenges in terms of filter capacities and flux rates. For these types of molecules, asymmetric PES parvovirus filter should be evaluated first, with and without prefiltration. If product recovery is an issue, regulatory agencies have accepted the use of non-parvovirus filters [2]. For smaller molecules, asymmetric PES parvovirus filters are recommended. Membrane-based prefilters could be used to normalize the feed (with respect to aggregate and impurity levels), increase capacity and reduce operating costs for this process step [2]. Special care should be taken when outlining the virus validation step as this will dictate achievable process loadings [2].

For the ultrafiltration/ diafiltration step, vendors typically recommend using filters 3-5X tighter than the molecular weight of the target molecule. Therefore, for Fab molecules that are typically small in comparison to Mabs (approximately 10kD – 80kD), 1-10 kD molecular weight cut-offs tangential flow devices are commonly used for the ultrafiltration/ diafiltration (UF/DF) step. As a result, the permeate fluxes can be lower. For BsAbs, the typical UF/ DF filters range from 30-50kD MWCO.

Additionally, some molecules can be PEG-ylated to improve bioavailability, which leads to higher viscosities as concentration increases. Also, because of the interest in subcutaneous applications, the target molecules are being concentrated to higher concentrations. For these reasons, the influence of the type of screen (screens help create turbulence and promote mass transfer) in flat sheet devices is paramount and should be taken into account in small scale optimization studies (1). A final consideration for E. coli expressed Fab molecules is downstream endotoxin removal. This is often achieved through anionic membrane adsorbers or charged membrane filters [1].

For the final sterile filtration, asymmetric PES membranes offer higher fluxes and capacities than symmetric membranes- however both PES and PVDF membranes should be evaluated at this stage. Attention should be paid to the final sterile filtration step as well, especially with respect to product recovery.

4.0 Part 3 – Scale Up and Process Transfer Considerations

Factors to consider when scaling-up current antibody and antibody variant processes depend on the stage and goals for the project, including the molecule’s pre-clinical or clinical phase, speed to market, process economics, manufacturing and operation flexibility, expertise, facility infrastructure, and batch volumes. Pros and cons of these factors can be weighted to decide how to proceed with the scale up logistics. In some cases, companies may lean towards utilizing single-use, stainless steel, or a hybrid for the manufacturing process. Additional considerations include building or using an existing facility, or outsourcing manufacturing to take advantage of the PD experience and infrastructure from contract manufacturing organizations (CMO).

Single-use processes have the inherent “out of the box, and ready to use” benefits, easing implementation.   With single use processes, users benefit from a lower investment in the upfront capital in comparison to a fixed facility with stainless steel systems, specific infrastructure requirements, such as steam and CIP/SIP, are likely not needed, and validation is minimal and/or eliminated [4]. For a multi-product facility, and in cases where batch volumes may vary and be less than 2000L, the additional benefits of a single-use approach may come from the quicker turnaround times from batch to batch, lower risk of cross product contamination, flexible volume manufacturing, and overall economics and facility fit. All of these factors contribute to the delivery of a process with speed to market needs, improved economics, and process flexibility.

A stainless steel facility can be considered for late stage molecules, multiple and large campaigns, and batch volumes greater than ~2000L where single-use systems may be a limitation.   In this case, the facility and equipment implementation would have a larger capital cost and initial validation investment; however, the long-term utilization of these assets can bring a return of investment and pay for itself over time, depreciation of equipment is incorporated, and other factors of a long term and multiuse facility may make economics more feasible [4]. A manufacturing facility may also incorporate a hybrid of single-use and stainless steel infrastructure to accommodate all needs of the project’s stage and goals. CMOs are well equipped with both types of facilities and process expertise, which may be more appealing in cases where there is a facility throughput limitation and/or speed to market may require outsourcing.

In addition to specific facility needs, each unit operation has its own specific rules for scale-up. In some cases, linear scalability can be accomplished for some technologies such as filtration; however, system scale-up is sometimes overlooked and can be the cause for deviation or unexpected process performance [5]. Fluid dynamics, hold-up volumes, frictional losses, hardware requirements and yield recoveries are factors to strongly investigate prior to scaling up. Ultimately, thorough process transfer studies must be completed to ensure the process meets specifications.

It is important to consider hold-up volumes of not only the devices utilized in the process, but also of the system itself and the impact this has on overall process recoveries. In some cases systems are installed in cramped spaces, which may require device selection and physical attributes of the tubing/piping to include turns and differential in height. Along with hold-up volumes and system/piping design all of which contribute to frictional losses, fluid dynamics (viscosity, temperature, flowrate is another factor to help determine the system component requirements for each unit operation [5]. In addition, when scaling up, researching the type of hardware to be used at large scale is evident, however, at times, not given enough attention. For example, there are many devices out on the market that are fully encapsulated at small scale, but for equivalent larger scales these may require holders. Considerations of the large scale hardware systems must be addressed within the different unit operations. Some of these include physical attributes, automation and footprint. Finally, yet important, proper validation of the systems and process should be completed prior to scaling up or even manufacturing of clinical material.

In addition to specific facility and system needs, all of the unit operations share a common ground of considerations for implementation and tech transfers. One of these considerations is for companies to further investigate each molecule’s process operating conditions via Design of Experiments (DoE) or even a deeper dive into a Quality by Design (QbD) approach. Another consideration is to understand raw material and consumable lot-to-lot variability, and the processes batch to batch variability. These factors can provide a better understanding of each unit operation and the performance of the process as a whole, which can contribute to the robustness and possible higher degree/window of operation. In cases where these deeper approaches may not be feasible, an upfront investment of rationally defined safety factors can be incorporated for all unit operations to minimize the risk for process deviations [6].

5.0 Conclusions
Antibody variants such as Fabs, BsAb and fusion proteins are generating increased interest as the demand for target therapeutics with improved efficacy continues to grow. Compared to traditional MAB processes, these molecules present some developing and manufacturing challenges. Each of the steps in recovery and purification of these molecules must be optimized based on the process requirements and the molecule characteristics, ensuring robust, stable and scalable production processes.

January 9, 2015 | Claire Scanlan | Mireille Deschamps | Juan Castano | Ruta Waghmare, PhD | Corresponding Author Ruta Waghmare , PhD | [email protected] | doi: 10.14229/jadc.2015.1.9.001

Received: December 10, 2014 | Accepted January 7, 2014 | Published online January 9, 2014

Creative Commons License