The Kibworth School, Kibworth, Leicester, England, UK
Received: February 25, 2019
Accepted: March 19, 2019
Version of Record Online: April 19, 2019
Blas P (2019) Improving the Bioprocessing of ADEPT Fusion Proteins using Ultra-Scale-Down Techniques. J Ann Bioeng 2019(1): 01-23.
Copyright © 2019 Peter Blas. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and work is properly cited.
Today with the advancement of modern medicine we are increasingly facing an aging community with a variety of complicated disease like the various forms of Cancer. Although man made synthetic chemicals can treat simple diseases, scientists predict that a new range of therapeutics containing antibody fragments and fusions proteins, may be important because of their ability to target specific sites. The new family of complicated biopharmaceuticals, such as fusions of larger molecules are more likely to be fragile and delicate to the unforgiving large-scale manufacturing environments. So new research needs to be conducted into where the damage may be occurring and how to stop it from happening.
This mini review focuses on the large-scale manufacture of antibody fusion proteins used to treat colorectal cancer with a novel drug delivery system called Antibody Directed Enzyme Prodrug Therapy (ADEPT). It explains how an Ultra Scale-Down (USD) shear device could be used to imitate the harsh process parameters using small amounts of protein solution. This would allow biopharmaceutical classification and recognition of new more efficient bioprocess parameters which could result in the enhancement of the large-scale process, improving yields and reducing unwanted impurities.
The information presented here reviews literature of the bioprocessing of antibody fusion proteins and explains that detailed understanding and characterisation in an ultra-scale-down shear device of small quantities of dilute protein solutions could be used to identify issues, like an amplified impurities levels in the biological product produced during the full scale GMP production.
Treating patients with diseases and illnesses in the modern era has required a new range of therapeutic products . These therapeutics are termed biologics or biopharmaceuticals and are typically recombinant proteins, vaccines, monoclonal Antibodies (mAb), fragments of antibodies, and fusions of proteins [2-4]. They are typically larger and more complicated than small synthetic chemical compounds and hence are particularly sensitive to loss in activity and function during bioprocessing, resulting in expensive production .
The problem is that large volume fermentations and downstream purification steps used in the manufacture of these complex proteins typically generate harsh conditions for a biomaterial to encounter . As a result, proteins can suffer from inactivation, degradation, aggregation and other forms of protein loss during bioprocessing. The causes of these losses could be a combination of the individual characteristics of the protein and/or the bioprocessing effects they encounter. Consequently, low process yields are experienced, which are thought to be due to the shear and secondary effects inflicted upon the protein during bioprocessing [7-9]. The biotechnology industries are under pressure to develop efficient manufacturing techniques that delivers high quality active products to the clinic in ever shorter periods of time.
A cost-effective way of improving large scale production can be the use of Ultra-Scale-Down (USD) technology . This is where millilitre quantities of precious process material can be exposed to bioprocessing conditions similar to those found during manufacture, resulting in the ability to generate effective large-scale working parameters [11,12]. Ultra-scale-down could be superior to pilot plant development work as it saves the process engineer time, manufacturing costs and is easier to operate .
The review focuses on the stability of a biopharmaceutical MFECP1 fusion protein intended for use in the treatment of colorectal cancer with a novel drug delivery system called Antibody Directed Enzyme Prodrug Therapy (ADEPT) , that has been in a phase 1 and 2 clinical trial. The therapy works by targeting an enzyme carboxypeptidase, CPG2, (42 kDa) to the tumour by virtue of its conjugation to a tumour specific antibody fragment, MFE (MFE is the single-chain Fv antibody molecule (27 kDa)). After sufficient time for circulatory clearance of this MFECP1 fusion protein a non-toxic prodrug is administered. This prodrug is converted to a highly cytotoxic drug by action of the enzyme at the tumour site [14,15]. The concern is to avoid degradation or denaturation of the complex during its bioprocessing.
In a competitive environment like antibody production the goal is to achieve efficient process development. This can be defined as “the ability to discover the optimum process parameters that produce the biopharmaceutical under pilot scale conditions in the shortest period of time” . The challenges and problems faced by process engineers at the pilot scale production stage of drug development may significantly increase the cost of these medicines by extending the time for development. It is hypothesised that scale down technologies can reduce these time lines, by identifying these problems earlier on in the process development stages and providing process solutions. Monoclonal antibodies and fusions of these proteins are thought to be the next generation of biopharmaceuticals; hence it is critical that a process engineer knows how these proteins function in high shear environments and how they are made.
It is thought millilitre quantities of this very high value process material can be characterised in an Ultra-Scale-Down (USD) shear device to identify ways to improve its large-scale bioprocessing . It can also be used to identify parameters that degrade the protein in defined shear conditions and uncover any important secondary shear associated effects. The review will seek to demonstrate that characterisation of process material in a USD shear device early on in drug development can help inform the large processing of biopharmaceuticals.
The mammalian immune system recognises and eliminates foreign pathogens by producing “antibodies” or immunoglobulins. These are specific glycoproteins that bind to foreign antigens. When a pathogen invades a mammal, an immunological response is triggered which involves the production of antibodies by plasma cells derived from B-lymphocytes. This automatic immune action helps remove invading parasites, viruses and bacterial toxins that may cause disease . Over time, this action is an adaptive process and therefore the immune system can improve with successive challenges by the same antigen. This is how immunity is created to diseases such as measles and diphtheria after an initial infection has been destroyed . The cellular immune system also uses killer T lymphocytes that recognise and destroy invading cells directly.
The unique ability of antibodies specifically to recognise and bind with high affinity to any type of antigen has made them interesting molecules for medical and scientific research . Potentially they could be used to target numerous illnesses if a distinctive antigen marker is identified. An antibody generates its high specificity by recognising distinctive amino-acid sequences on an antigen called an epitope and/or distinctive features on the surface of other types of antigens including polysaccharides. Antibodies are believed to be the most important part of the immune system . Also, from a financial point of view they are particularly important as 25% of pharmacological agents recently under development were based on antibodies  and this class of therapeutics was growing at a fast rate .
Polyclonal and monoclonal antibodies
Initially polyclonal antibodies were produced from small mammals such as mice, rabbits and goats by inoculating them with an antigen which generated the production of different antibodies. These different antibodies (polyclonals) recognised multiple epitopes on different antigens resulting in poor specificity. A range of foreign molecules injected into a mammal will elicit a response; antigen proteins from different species can be injected into another animal to generate polyclonal antibodies against that antigen protein. For example, a human protein can be put into a mouse to generate mouse (murine) antibodies. Initially animals were inoculated and bled for their antibodies. Hence the larger the animal was, the greater the volume of antiserum that could be extracted.
Serum obtained from an animal inoculated with an antigen is a good source of polyclonal antibodies and is often used in the development of immunoassay techniques. Earlier work in this area generated polyclonal antibodies from rodents treated with a foreign antigen that affected humans. Successful work with polyclonal antibodies was carried out in the treatment of snake bite victims. However, most crude preparations of polyclonal antibodies were ineffective because they produced an immune response in humans .
Georges Köhler and César Milstein recognised the potential advantage of antibodies and as a result of their work produced hybridoma technology winning them the 1984 Nobel Prize in Physiology and Medicine . A hybridoma cell is created by fusing an antibody producing B-cell from an immunised rodent with a myeloma tumour cell. The resultant hybridoma cells can be cloned to produce exactly the same murine antibodies indefinitely in cell cultures. Therefore, monoclonal antibodies are homogeneous in specificity, which means they have identical binding regions that attach to the antigen used to generate them in the first place.
Structure of antibodies
In vertebrates there are five classes of immunoglobulins, IgG, IgM, IgA, IgD and IgE, which differ by their function in the immune system . From this group IgGs are the most common and make up 70-75% of the total immunoglobulin pool in the bloodstream. All IgGs have a standard structure consisting of two identical Heavy Chains (H) and two identical Light (L) polypeptide chains linked together by disulfide bridges and non-covalent interactions . These chains form a Y shaped homodimer; each heavy polypeptide chain consists of two 450 amino-acids and each light chain consists of two 250 amino acids. The whole IgG molecule (Figure 1) has a mass of around 160 kDa. The binding specificity of an antibody is generated in the variable regions VH and VL, which are located at the N terminal of the molecule . Each variable region contains three loops connecting the β-sheets together. These loops exist in a wide variety of lengths and sequences generating different binding regions between antibodies. Hence in total there are six hypervariable loops, known as the complementary determining regions, which can bind to a range of antigens . The random nature of the amino acid sequences in these variable regions results in the antibody having the ability to bind to a diverse array of antigens. The actual binding area at the variable region is called the paratope. This binds to an epitope which is the unique section of amino acid residues on the antigen (Figure 1) [24,25].
Figure 1: The figure above shows the conventional Y shaped structure of an Immunoglobulin (IgG) antibody molecule. The antibody is comprised of two Heavy chains (H) which contain longer amino acid sequences and two Light chains (L) which contain shorter ones. The variable areas are coloured red and orange and are a combination of light and heavy chains. The structure can be cleaved in places to generate Fc, Fab and Fv. These fragments are then used in the production of fusion proteins used in ADEPT .
In addition to the variable regions, each L chain contains one constant region (CL) and each H chain contains three constant regions CH1, CH2 and CH3. The amino acid sequences of these remaining C-terminal domains (constant regions) are much less variable. The non-variable structure remains consistent throughout a range of antibody classes and plays further roles in the immune system. These regions are responsible for interacting with phagocytes and activating processes that attack pathogenic toxins. Over all the whole molecule works as a key part of the immune system helping eliminate specific pathogens and toxins entering the body .
Types of fragments
Cleavage of the disulfide bonds that hold an antibody together can generate two Antibody Binding Fragments (Fab), and one Crystallisable Antibody Fragment (Fc), these fragments can be seen in figure 2. Each Fab fragment is composed of the L chain and the N-terminal of the H chain, and is monovalent. It is therefore able to bind to just one epitope. The Fc fragment readily interacts with Fc receptors on cells and is comprised of the CH2 and CH3 domains. Fab fragments are homodimers, which means they have two similar domains bound by non-covalent bonds, consisting of the VH, CH1, CL and VL domains and are usually of molecular mass around 50 kDa . The Fc fragment is a non-specific binding domain and is a constant region found in all antibody classes. This Fc fragment is normally cleaved or not even used in industry as it has a very low binding affinity to various target antigens. On the other hand, Fab fragments encompass all the major binding domains used in the non-covalent interaction with antigens. Treating a whole IgG molecule with a non-specific enzyme like papain, a digestive protease, which breaks the bonds holding the heavy and light chains together, can produce Fab and Fv fragments. Pepsin a specific enzyme can cleave away just the Fc region from a whole IgG molecule, generating a bivalent fragment with the ability to bind to two molecules of the same antigen .
Figure 2: The figure above shows diagrams of antibody fragments available after enzymatic digestion of the whole intact immunoglobulin molecule. The fragments above included, Fab, Fv and ScFv. The Figure shows that the VH and VL domains of the Fv can be linked in different ways, with different numbers of linker amino acids. All of the above fragments are used in drug delivery systems and fusion proteins, recognising specific antigen on target cells to treat different illnesses. The figure shows that these fragments could be fragile to shear associated effects found during the bioprocessing of these proteins .
An Fv fragment is around 30 kDa in size and contains the Heavy (VH) and Light Chain (VL) domains. It is the smallest part of the antibody that has the ability to bind to an antigen . Native Fv fragments are generally unstable at low protein concentrations as they are not bonded together . Due to the importance of this targeting agent in medicine, industry has found two ways to overcome the degradation of the VH and VL domains. Firstly, these Fv domains can be joined by a peptide linker, for example a Gly4 Ser3 peptide, producing a structure called a Single Chain Fv (ScFv) . Longer linkers of around 15 amino acid residues can be used if flexibility is required in the design of the drug and/or to enhance solubility. Another strategy is to generate disulphide bonds between the domains by the reduction of cysteine residues producing a structure called a Stabilised Fv (dsFv) . Fragment antibodies are advantageous over whole antibodies as they have good clearance from the body and fast tumour penetration properties , also they are not glycosylated, and therefore can be made in a variety of expression systems.
Application and function of antibody fragments: Monoclonal antibodies have great power as they have homogeneous specificity against antigenic proteins associated with infectious diseases. Antibody fragments, like Fabs and ScFv, hold even greater advantages because when it comes to therapeutic application their reduced size allows them to readily penetrate tissue and solid tumours. These fragments are also well suited for medicinal imaging, treatment of over dose and targeting tumours .
Initially antibodies and their corresponding fragments were thought to be the “Magic Bullets” of the new scientific age, with the promise to treat a host of diseases. It was thought antibody-based therapies could put an end to unwanted side effects generated by non-specific targeting of toxic drugs to healthy tissues. As a result of this potential advantage, immunoagents became popular for the treatment and diagnosis of cancers [4,15]. A new and exciting approach to treating disease is the use of antibody fragments attached or fused to toxic isotopes and proteins. This allows the direct delivery of protein drug to the site where it is required . An example of an approved monoclonal antibody to treat cancer is the drug marketed under the name HerceptinTM by Genentech, which stops the spread of breast cancer by blocking growth receptors on infected cells. Clinical trials with Herceptin and traditional chemotherapy increased the life expectancies of women with advanced forms of the cancer by five months . Another novel therapeutic antibody ZevalinTM was approved in 2002 by the Food and Drug Administration (FDA) for the treatment of non-hodgkins lymphoma.
Humanisation of antibodies: Researchers soon realised that murine monoclonal antibodies still proved to be ineffective in the clinic because they elicit a strong immunogenic response when injected into humans. This is because these antibodies were 100% mouse proteins and therefore considered foreign by the human immune system. The injection of therapeutic murine antibodies to treat disease in humans generated the production of Human Anti-Mouse Antibodies (HAMA). These have the ability to clear rapidly foreign proteins from the body. This HAMA response caused most early monoclonal antibodies to be ineffective . Further difficulties with murine antibodies included their inability to activate macrophage cells (white blood cells) that were supposed to destroy the antibody-antigen complex. It was found that differences in the heavy and light chains between mouse and human antibodies were the reasons why these critical effecter cells were not activated. Despite these major drawbacks, technology progressed and some success stories were reported, one being the manufacture of Orthoclone (OKT3). This was approved in 1986 to help prevent the rejection of transplanted organs .
Further research turned to the area of genetic engineering in an attempt to reduce the HAMA response. Genes from mouse antibodies were grafted on to the genes for human antibodies. This action produced what is known as a chimeric antibody which had 70% human domains and 30% of the mouse variable domains [32,33]. These antibodies have good affinity for the target antigen and produce a lower HAMA response. Successful chimeric antibodies include ReoproTM made by Eli Lilly and approved in 1994 to treat blood clots in patients with cardiovascular disease, RituxamTM produced by Genentech, approved in 1997 to treat non-hodgkins lymphoma and SimulectTM produced by Novartis, approved in 1998 for the treatment of transplant rejection. Although these were successful, problems still persisted with the antigenic immune response and it was thought that a totally humanised antibody would be more effective.
Pioneering work by Greg Winter at Cambridge University in 1986 reduced the mouse component in a chimeric antibody down to 5-10% . The important regions on an antibody are the 3 β-loops in the variable region of the antibody arm; the rest can be substituted with human antibody genes without destroying the strong antigen binding qualities. The resulting “Humanised antibodies” produced a much lower antigenic response and proved to be very successful . This work led to a rapid growth of humanised monoclonal antibodies in the biotech industry and in 1997 Roche produced the first humanised monoclonal antibody “ZenapaxTM” which was used to combat organ rejection . Since the work by Winter, the industry has enjoyed a healthy growth in humanised monoclonal antibodies that have been approved by the FDA.
Specificity of antibody fragments: The unique specificity of antibody fragments is generated from B-lymphocytes, as they use a process known as somatic mutation that randomly changes the sequences of amino acids in the hypervariable binding region. This very powerful process helps optimise the binding of antibody fragments to foreign antigens, as the most favourable amino acid combination is expressed depending on the type of antigen . The binding action of antibodies does not involve the direct covalent linkage between antibody and antigenic protein. Instead, varying strengths of van der Waals, hydrogen bonding and hydrophobic attractions are involved in the antibody/antigen interaction. All of these contacts are generated by differences in amino acid residues of the hypervariable regions. These forces are considerably weaker than covalent bonds; however together they form a considerable effect . Electron clouds of atoms in different amino acid side chains form the interactions between paratope on the hypervariable region of the antibody and the epitope on the antigen. If, however these clouds do not fit in a lock and key type way and very strong repulsive forces prevail, then binding is weak . The sum of the attractive and repulsive forces generated by the electron clouds of the atoms is a good measure of the specificity or binding affinity of the antibody. Another measure of antibody binding strength is known as avidity which is defined as the combined strength of multiple bond interactions, i.e. as the sum of the binding strength of all domains present in the hypervariable region.
Antibody fusion proteins used in therapy
The unique and very specific antibody antigen interaction allows the targeting of drugs to infected or diseased tissues. Fusions of antibody fragments attached to other proteins can be used to help treat disease [4,14] and/or aid in the imaging of cancerous areas of the body . ScFv antibody fragments which have high affinity for cancer antigen markers are normally used because they can diffuse quickly into tumours due to their relatively small size. Radioimmunotherapy (RIT) and Antibody Directed Enzyme Prodrug Therapy (ADEPT) are drug delivery systems that use antibodies fused to other molecules to combat complex diseases like cancer. RIT works by targeting a radioisotope, normally I131, that is conjugated to a ScFv antibody fragment to cancerous cells that have a distinctive antigenic surface marker . This type of specific drug discovery process is normally used to treat lymphoma cancers. Other examples of recombinant fusion proteins include Etanercept, (EnberlTM) produced by Amgen, which was used to treat rheumatoid arthritis. The following sections will review areas of ADEPT which specifically use complicated antibody fusion proteins.
Antibody Directed Enzyme Prodrug Therapy
Antibody-Directed Enzyme-Prodrug Therapy (ADEPT) is a complex two-phase drug delivery system, in which an enzyme is targeted to a tumour site by virtue of its attachment to an antibody, where it selectively activates a relatively non-toxic prodrug into a potent cytotoxic agent . This technique was first proposed by  and serves as an alternative to antibody guided radiation (RIT) or other radioimmuno-conjugates because of its superior tumour uptake ability.
The first part of the system works by administering the recipient with a MFECP1 fusion protein, which consists of a ScFv antibody fragment (designated MFE-23) fused to an active enzyme, in this case carboxypeptidase (CPG2) . Next the ScFv antibody fragment recognises and attaches to a cancer glycoprotein Carcino-Embryonic Antigen (CEA)  which is highly expressed on the surface of cancer cells in patients with colorectal carcinoma .
The unique affinity of the antibody fragment to the cancer antigen marker CEA allows the accumulation of the MFECP1 fusion protein specifically to the site of action. After sufficient time for clearance of unbound antibody fragments from the rest of the body, a relatively harmless nitrogen mustard prodrug is administered. This prodrug is cleaved into a cytotoxic drug by the CPG2 enzyme, at the tumour site resulting in the death of cancerous cells. The enzyme CPG2 is able to catalyse many molecules of prodrug into the active toxic molecules at the site of action; therefore, this process becomes very efficient and is not directly dependent upon the amount of MFECP1 fusion protein administered.
Prodrugs in ADEPT
The active drug has a low molecular weight so it can easily diffuse through the tumour producing a bystander effect . Different prodrugs vary in efficacy but it has been found that more reactive nitrogen mustard compounds produce shorter half lives in the body, and are more potent for ADEPT. The most commonly used and studied prodrugs are the nitrogen mustard group of compounds that are deactivated into prodrugs before administration. See figure 3 for chemical structures of prodrugs. The Pseudomonas carboxypeptidase G2 (CPG2) was considered to be an ideal enzyme to be used in ADEPT as it hydrolyses a glutamate moiety from the nitrogen mustard prodrugs causing them to become cytotoxic drugs .
Figure 3: Above shows the enzyme and prodrug used in Antibody Directed Enzyme Prodrug Therapy (ADPET). (3a): Above shows a picture of carboxypeptidase CPG2 in the homodimer complex with two zinc ions in the active site showing how specific it is at cleaving the carbonyl group; (3b): Shows the chemical structures of prodrug (ZD2767P) and cytotoxic drug (ZD2767D) it forms when the CPG2 cleaves it, this cytotoxic drug then destroys the tumour cells specially at the site of action. If the enzyme is denatured or sheared during the bioprocessing in anyway this action cannot happen rendering ADEPT ineffective .
Studies have shown that CPG2 fusion proteins in combination with a nitrogen mustard prodrug have been successful in reducing solid tumour size in animal xenograph models and human cell lines [14,37]. Further to this, encouraging effects of the fusion protein MFE-23::CPG2 with nitrogen mustard prodrugs in human patients with colorectal cancer have been reported . These investigations showed that fusion proteins have the potential to treat specific cancers by utilising prodrug activation with very high therapeutic efficacy .
ADEPT has several advantages that make it a useful system in combating specific cancers. The first is the bystander effect as the active drug is a small molecule that can actively diffuse through the tumour to reach antigen positive cells . Each CPG2 enzyme on the MFECP1 fusion protein can catalyse several prodrug molecules into active cytotoxic entities, which acts as an amplification step. Also, systemic toxicity is reduced as the active drug is only produced at the cancer site . However, the draw backs include the need to maintain the MFECP1 fusion protein in the body for prolonged periods of time. This effect can produce immunogenicity problems and could reduce the efficiency of MFECP1 fusion protein uptake by the tumour due to antibodies raised against the protein during repeat doses .
Enzymes in ADEPT (Carboxypeptidase CPG2)
Carboxypeptidase CPG2 is a bacterial enzyme used in ADEPT and various other cancer therapies. This enzyme specifically hydrolyses the C-terminal glutamate moiety from folic acid and various other analogues such as methotrexate  (Figure 3). Folates are essential coenzymes used in the synthesis of Deoxyribonucleic Acid (DNA) via pyrimidines and purines. Analogous compounds such as methotrexate are used to treat tumour cells because they inhibit enzymes, such as dihydrofolate reductase, that help construct DNA bases . Therefore, CPG2 is used in other rescue cancer therapies to eliminate excess methotrexate by breaking it down into a non-toxic entity, necessary because of the severe side effects excess methotrexate can cause in the body . It must be noted that methotrexate rescue therapy is a very critical clinical situation, as high amounts of this drug can cause toxic effects on rapidly dividing cells of gastrointestinal mucosa and bone marrow.
The actual CPG2 enzyme is a dimeric zinc-dependent exopeptidase which can be produced in Pseudomonas sp. strain RS-16 , Escherichia coli (E.coli)  and Pichia pastoris . The enzyme relies on the presence of two coenzymes, which in this case are the zinc metal ions for effective catalytic activity to occur. The dimer consists of two domains in each sub-unit; the first domain forms a β sheet which provides the interactions at the dimer interface, while the second domain forms the catalytic centre . The catalytic domain contains two zinc ions tetrahedrally bound at the active site with carboxylate oxygens (which are oxygen atoms that are double bonded to a carbon atom). Each zinc ion binds via one histidine, one glutamate and one aspartate amino acid side chain ligand, where one aspartate and a water molecule act as bridging ligands . The existence of the homodimer, means that each individual enzyme is non-covalently linked in pairs via the β-sheets (Figure 4).
Figure 4: Above shows a 3-D diagram of the CPG2 enzyme and where the prodrug is activated by the enzyme. The monomers in red and grey join to form a dimer (red and grey complex above). The catalytic centres contain a potential shear fragile zinc metal ion region. This diagram shows that the fusion protein in ADEPT can exist as dimer form and this might affect the stability during the large-scale bioprocessing of the protein.
Source: Royal Free Hospital, 2004.
CPG2 catalyses the conversion of inactive prodrugs, (so called Nitrogen mustard prodrugs) into active cytotoxic agents which destroy cancerous cells. The CPG2 enzyme is an efficient enzyme which is why it is used in ADEPT as an activating step to treat colon cancer.
Discovery of antibodies in ADEPT
Antibodies and antibody fragments are able to target cancer cells selectively by virtue of the unique binding affinity between antibody and antigen, this process can be used to deliver therapeutics and imaging agents to tumours . Initially specific antibody fragments were used to locate the tumours, with radioactivie Iodine123 isotope labelling. However most recently they have been used to direct cytotoxic entities to tumours .
Phage display is the technique used to generate libraries of ScFv fragment antibodies with a diverse array of binding abilities. The first step is to insert diverse genetic material into a phage genome. Next the bacteriophage processes this new gene so that a new protein is expressed on the phage surface. The surface of each bacteriophage displays the antibody encoded by the gene construct that it contains. Bacteriophages with the desired antigen are readily selected from libraries of repeated rounds of binding to immobilised antigens . Clones to create ScFv fragments from the library with the highest affinity for the antigen could be found. This is how the MFE-23, the ScFv fragment used in ADEPT, was discovered [15,40].
MFECP1 fusion protein structure
Figure 5, shows a picture of the whole intact MFECP1 fusion protein used in ADEPT. Blue sections show the MFE-23 ScFv antibody fragment which has affinity for the cancer antigen CEA. The red structure shows the CPG2 which hydrolyses the prodrug into a cytotoxic entity. The purple section shows the hexa-histidine tag used for purification of the MFECP1 fusion protein by IMAC from fermentation broth. Yellow sections show the mannose sugars (glycosylation) specifically from Pichia pastoris. The CPG2 enzyme is around 42 kDa and the MFE-23 antibody fragment is around 28 kDa together with leader sequences and mannose sugar glycosylation give an approximate size of the MFECP1 fusion protein to be around 70 kDa. Figure 6 shows how the MFECP1 fusion protein may exist in its natural state as a homodimer as the CPG2 enzyme readily forms this structure.
Figure 5: Above shows a picture of the whole intact MFECP1 fusion protein 70 (kDa) used in ADEPT. Blue sections show the MFE-23 ScFv antibody fragment which has affinity for the cancer antigen CEA. The red structure shows the CPG2 which hydrolyses the prodrug. The purple section shows the hexa-histidine tag used for purification by IMAC. Yellow sections show the mannose sugars (glycosylation) from pichia pastoris. This figure shows the possible fragments that could be produced during the large-scale bioprocessing of the fusion protein.
Figure 6: Above shows how the MFECP1 fusion protein may exist in its natural state as carboxypeptidase CPG2 exists as a homodimer therefore a complex dimer is formed. This diagram shows that potentially this fusion protein could be susceptible to various shear forces causing degradation, deactivation and denaturing during the large-scale production and purification stages.
Recombinant proteins are usually produced in high cell density fermentation cultures using expression systems selected to yield high protein titres. The fermentation process is normally carried out in bioreactors which are vessels that allow cells to grow under controlled physiological conditions so that they can express the protein of interest. The most important conditions to maintain for the cultures to survive are Temperature (T), dissolved Oxygen Concentration (DO), pH, nutrient concentration (e.g. carbon sources) and mixing. The production initially is scaled up from shake flask cultures (approximately 0.250 L) and eventually ends up at the fermentation scale (which can be 10-100 L or more) depending on how much protein is required. The protein product which resides in the fermentation broth is then concentrated and purified in downstream processing steps.
The type of downstream processing steps taken depends on where the protein of interest is compartmentalised. Proteins expressed intracellularly or in the periplasm of the cell require cell lysis for their release while cells that secrete proteins do not require breakage. Downstream processing can vary from product to product, but essentially should aim to remove all cells and debris from the process broth, leaving behind a low volume, highly concentrated protein solution. Buffer exchange and concentration steps are normal practice. During production, regulatory requirements must be followed and these may include essential buffer exchange or pH holding steps for virus inactivation. The types of expression system, fermentation and downstream purification steps employed for a particular recombinant protein depend on many factors, some of which include, cost, capabilities of the equipment, time and characteristics of the product. These elements and others will be described fully in the following sections.
Expression systems selection
A good expression system should produce the recombinant protein of interest in the most cost-effective way, with ease of operation and in the fastest possible time. The types of expression systems used to produce a therapeutic protein are very important as each can modify the product differently, giving for example different glycosylated variants or different protein configurations (e.g. tertiary or quaternary structures). The major factors that should be considered when choosing an expression system are productivity, ease of operation, knowledge of the cell line, time frames required, scalability, product characteristics, regulatory issues and cost.
A variety of host cell expression systems are used in industry to produce a range of proteins, some of which include bacteria, yeast, mammalian, fungal, insect, and plant cells. Primarily bacteria like Escherichia coli. was initially used for protein expression as they were simple prokaryote organisms that had been well characterised [41,42]. However, it later became apparent that prokaryotes are unable to produce certain functional proteins such as recombinant human tPA, Factor VIII, erythropoietin and monoclonal antibodies as they are unable to fold proteins correctly and offer the right post-translational modifications . These problems pushed research into mammalian cell cultures  where more complex products could be grown with sophisticated cell lines such as Chinese Hamster Ovary (CHO) . Unfortunately, mammalian cell cultures are particularly sensitive to shear stresses in a bioreactor, resulting in lower cell titres/yields .
Yeast cultures on the other hand are much more robust than mammalian cells and are able to perform some of the complex folding and post-translation modification required for some mammalian protein products. It was found that different host cell expression systems expressed a range of protein titres, reflecting their shear sensitivity. A general illustration of the relative fragility of various host cell lines can be found in  and can be seen in figure 7.
Figure 7: Above shows estimations of the range of stress in strength N.m-1 that may be required to cause cell breakage for a range of cell types. The figure shows that various types of cells have varying levels of shear sensitivity, plants cells being the most robust and mammalian cells being the most fragile. This is important as disrupted cells can release various proteases and enzymes into the process stream which can lead to fusion protein destruction .
Even though recombinant technology was first employed in the 1970s to generate insulin , scientists today still strive to improve the yield obtained by a particular host. Work by Plantz et al.,  improved the amount of a recombinant ovine interferon-t product in yeast Pichia pastoris fermentation by optimising productivity and reformulation of growth medium. A robust and stable fermentation process was devised that increased production concentrations from 203 to 337 mg/L and increased the initial process yield resulting in an overall 210% increase of total yield from 557 to 1,172 mg .
Lin et al.,  investigated the effects of salt supplementation, batch glycerol, pH and temperature on production of an Fc fusion protein in Pichia pastoris cells. The pH was found to be particularly critical, pH 7.2 was the optimum and improved the Fc protein titre yields from 291 mg/L to 373 mg/L . However, more importantly it was found that optimum fermentation conditions for Pichia pastoris were protein specific and there were no set pre-defined conditions that could work well for all type of proteins . It is hypothesised by this group that complex fermentation development for each recombinant protein will be required in order to achieve good yields in large scale production .
Pichia pastoris: Pichia pastoris is one of the four methylotrophic yeast genera which can be used to express recombinant proteins for research and industrial purposes [49-51]. The system is based on the extremely strong and tightly regulated promoter from the Alcohol Oxidase I Gene (AOX1), that is used to drive the expression of the foreign gene . Advantages that make this cell line the premiere choice for researchers include; 1, the yeast is well characterised; 2, it can perform higher eukaryotic protein modifications such as disulfide bond formation, advanced protein folding and glycosylation ; 3, it can produce high cell density fermentations with good titres ; 4, it can be manipulated at the molecular level hence, foreign genes can be introduced easily ; 5, it can also use methanol as a sole carbon source. Additionally, it is more robust than mammalian cell lines and fermentations are less costly. The culture media and recipes describing how Pichia pastoris can be used to produce high cell density fermentation have not changed much from those originally developed by the Phillips Petroleum Company in the 1970s  and are much cheaper than mammalian cell culture recipes.
There are some disadvantages associated with recombinant protein production in Pichia pastoris. Glycerol is required for initial cell growth; however, even relatively slow glycerol feed rates can cause severe repression of recombinant protein synthesis . Furthermore, glycerol metabolism can trigger the build-up of acetate and ethanol to levels that can repress the AOX1 promoter, consequently decreasing foreign protein production . Other disadvantages include the excessive requirement of methanol which at 100-200 L scale can present a significant health and safety problem due to the possibility of explosions. Recombinant proteins can be expressed into different compartments of the cell; expression into the cytoplasm and periplasm; is known as intracellular expression and secretion outside the cell is known as extracellular expression. The leader sequence encoded before the gene vector governs where the protein will finally reside.
Fermentation as referred to in this thesis refers to the growth of a biological expression system with critical raw materials in aerobic conditions to produce the protein of interest. The protein can be produced intra or extracellularly and depending on the vector that encodes for the protein. Traditionally, microbial fermentations are carried out in mechanically driven stirred tank bioreactors which typically have a 3:1 aspect ratio and are made of stainless steel (319SS) if larger than 20 L, while smaller vessels are sometimes made from glass [53,54].
The vessels maintain the pH, temperature (via cooling jacket) and dissolved oxygen of the culture to ensure the microbial cells grow effectively. The agitation is normally provided by two to three Rushton turbine impellers each with six blades that generate a homogeneous environment throughout the culture and promotes adequate bubble dispersion. Four metal baffles are fixed to the inside of the tank to stop the culture medium from vortexing and also ensuring turbulent flow for good oxygen transfer. Aeration with air/or oxygen is achieved via a gas sparger at the bottom of the reactor below the impellers. Air exiting the vessel is filtered by High Efficiency Particulate Air (HEPA) filters to retain airborne microbes. Cultures are grown for specific lengths of time after which the protein to be expressed is induced to start transcription if a recombinant product is to be produced, however this is not necessarily true for all fermentations. After protein expression, the product is ready to be purified by different downstream processing techniques (figure 8 shows a diagram of a typical fermenter with attached inlet/outlet equipment).
Figure 8: Above shows a picture of a typical fermenter used in recombinant protein production. The figure shows all of the key pieces of equipment required to maintain and produce a recombinant protein.
Downstream Processing (DSP) essentially involves the capture and purification of the protein product of interest and the removal of all major contaminants arising from the fermentation culture. The function of downstream processing is to generate a final (protein) product of very high purity (ideally >99%) free of protein and microbial contaminants that, in its final formulation, meets the stringent requirements of regulatory agencies (such as FDA or the Medicines and Healthcare products Regulatory Agency (MHRA)) for efficacy, consistency and purity.
A range of factors will influence the types of unit operations to be used in the purification of the protein in question, including the compartmentalisation of the protein (intra or extracellular), associated ligands/physiochemical properties of the protein, the cost of the process, the final desired purity, the amount of the desired product required, the ability to scale-up the process to a larger scale and any further rules imposed by the regulatory authorities. It is also essential to verify that the protein in question retains its quaternary structure with all fragments and fused areas of the protein intact for effective therapeutic efficacy and activity. This last requirement may restrict the use of some potentially damaging methods or force the use of a more expensive stage like chromatography. It has been calculated that 80-50% of the production cost for new biopharmaceuticals comes from the downstream purification route taken . It is therefore essential that the most efficient steps are taken to meet the required targets. Each different unit operation taken in the downstream processing results in some product loss and these become cumulative over several steps. For example, Doran PM  explains that an 80% recovery of a product for each of five downstream processing steps results in an overall product recovery of only 33%.
The downstream purification of culture medium has distinct goals that must be followed, these include; the release of the protein product into the supernatant, the removal of bulk cellular debris and the removal of water to concentrate the protein. Sequential unit operations are used that utilise different physiochemical properties of the protein to generate the correct purification train . Initially a capture step, for example expanded bed absorption, centrifugation, cell lysis and micro-filtration, is used to separate the protein product from the cells. This first step removes a majority of the cellular matter and cell components depending on where the protein resides. Intracellular proteins require cell breakage before capture can begin; extracellular proteins are slightly faster to process, as the cells can be separated from the supernatant easily. After precipitation of microbial cells, chromatography can be used to isolate the protein from the supernatant . Most contaminants are removed by different forms of chromatography followed by a concentration/diafiltration and polishing stages . Concentration/diafiltration reduces the handling volume and can allow buffer exchange. Polishing steps are the last section of DSP and normally involve size exclusion chromatography to remove contaminating breakdown products of the protein present. Endotoxin removal is a compulsory requirement enforced by the regulatory authorities, these toxic cell by-products can be removed with charged columns.
The following sections will review the types of downstream unit operations required specifically in the capture and purification of the MFECP1 fusion protein under investigation.
Expanded Bed Adsorption
Expanded Bed Adsorption (EBA) is a harvesting technique that allows product capture and concentration of the fermentation broth . A specially designed porous matrix with a product binding ligand forms the stationary phase in a dynamic fluidised bed. Buffer is pumped through the column expanding the porous bed height to approximately five times its original height. After the bed has expanded, unclarified culture broth, which normally has been conditioned (e.g. diluted with high salt solutions) to allow ligand binding, is pumped through the bottom. This action allows the cells to pass through the matrix gaps and protein to bind. After all the broth has been passed through the column, the expanded bed is compressed to its original size by reverse direction of flow and the product is eluted with an appropriate buffer that dissociates the protein from its ligand. The compression of the bed allows product concentration in a small volume, which can then be easily handled and moved to the next purification stage. Immobilised Metal Affinity Chromatography (IMAC) is a sophisticated type of EBA which has a transition metal ion, Cu2+ or Ni2+, attached to a chelating porous matrix particle. This metal ion has a very specific and high affinity for a hexa-histidine-tag which is often encoded as part of the protein amino acid sequence to facilitate specific capture of the protein from the un-clarified fermentation broth.
There are many advantages of using EBA, as it can potentially cover two-unit operations, thereby reducing the potential for product loss. However, there are disadvantages; the capture matrices may be very expensive and for re-use, the column must be cleaned effectively, using a validated process; such matrices often have a limited life span and exhibit loss of absorption performance resulting from the use of harsh cleaning regimes [39,58,59].
Tangential Flow Filtration (TFF)
This is different from conventional filtration as the process fluid is passed tangentially across a filter membrane instead of directly through it. Tangential flow filtration systems can allow the concentration and buffer exchange (dia-filtration) of the process stream by a recycling process. TFF has the advantage of drastically reducing the fouling times and hence development of cakes on the filter [60,61]. A positive transmembrane pressure achieves permeate flux through a filter; however, the higher the pressure the faster the membrane will foul over time. Therefore, in development it is critical to have low transmembrane pressure and high permeate flux rates. TFF can be used to concentrate the product up stream of the filter or to separate the product from larger molecular weight contaminants by allowing the product through the filter. Product selection from the process stream can be achieved by selecting membranes with appropriate molecular size cut offs.
The bioprocess industry relies on specific downstream chromatographic steps to separate and purify protein products from the process streams . The protein solution entering the column with appropriate buffers is called the mobile phase and the porous packed bed of matrix in the column is called the stationary phase. The interaction between the mobile phase and the stationary phase allows separation of contaminants and product. Selective stationary phases with appropriate physiochemical properties and/or ligands can selectively capture species of interest. Afterwards using suitable elution strategies, for example by changing the mobile phase concentrations, can cause selective elution of the product of interest. Elution gradients can be used to change the retention times of the solutes which can be helpful when removing contaminants. Homogeneous packing of the stationary phase in the column is critical to obtain high resolution and reproducible results. Any gaps or air bubbles in the column bed will hinder the chromatographic separation. Various chromatographic techniques are used to capture and purify protein solutions; these include, size exclusion, affinity, hydrophobic interaction and ion exchange . Size exclusion, Fast Purification Liquid Chromatography (FPLC) was used to capture and purify the MFECP1 fusion protein and is explained in the next sections.
Size exclusion chromatography: Size exclusion chromatography (a type of FPLC) does not use weak ionic interactions between the product in the mobile phase and the stationary phase to separate species. Instead it uses steric hindrance and size differences between products in the mobile phase to separate out impurities on the basis of molecular weight. Most of the time this technique is used as a final polishing step as it removes aggregates and small fragments associated with the protein . This chromatography works by excluding large molecules to a greater extent than small molecules from the compressible gel matrix stationary phase in the column. This action allows large proteins to migrate faster through the column as they by-pass the matrix and come out first. However small proteins enter the gel matrix holes and hence access a greater proportion of the column volume and take longer to elute. Size exclusion columns are generally longer and thinner than affinity, ionic and HIC columns as the resolution largely depend on the length of the column. Also, they are run at much slower flow rates than other chromatography columns as resolution between smaller molecular proteins can be compromised if the flow rate is too high . In addition to this other disadvantage of this type of chromatography can be that it dilutes the pure protein product.
The regulatory authorities insist that all biopharmaceutical products for patients must have steps in the production processes to remove potential toxins, for example viral contaminants and bacterial endotoxins. This is to ensure patient safety, as recombinant proteins by virtue of their production and origin could easily become contaminated.
Endotoxins are Lipopolysaccharides (LPS) of the outer cell wall of Gram-negative bacteria . They produce strong toxic effects in human and animals, even at very low concentrations. These effects include changing the structure and function of cells, changing metabolic functions, raising body temperature, triggering the coagulation cascade and causing shock . Hence because of these effects it is critical to remove any potential endotoxins that may be in the biological product, especially if it is an injectable therapeutic. Fortunately, endotoxins contain polar hetero-saccharide chains that can be removed by passing them through an affinity column under gravity. Affinity ligands, like the amino acids arginine or serine, can be used to remove nearly 78 % of all endotoxins . The solution is normally passed through the columns several times, to reduce the concentration of endotoxins to an acceptable level.
There are two ways in which viruses can contaminate a biopharmaceutical product. Contamination may derive from the original animal cell line or be introduced by the addition of medium components of animal origin. Regulatory bodies require that two viral inactivation steps are incorporated into the process of an approved drug. According to EU regulations, at least one of the inactivating or removing steps should be effective against Non-Lipid Enveloped (NLE) viruses . Viral inactivation requires the addition of harsh chemical conditions, heat, and/or UV irradiation. With this in mind it is important that the engineer does not damage the product in anyway. Low pH acid holds are regularly used after a Protein A elution; treated with detergents or urea can also be used. Affinity chromatography, depth filters, ultrafiltration membranes and tangential flow filtration can all be used to capture viral contaminants.
Proteins are fragile biological entities that are very susceptible to changes in the external environment due to their aliphatic nature. They contain hydrophobic and hydrophilic areas within their structure, generated by virtue of the different charged amino acid sequences, which are susceptible to conformational changes . The differences in amino acid sequences also generate many different shaped domains and sizes, which could affect their potential bioprocessing. When biological proteins are handled at low shear stresses, they generally stay in their native conformational forms . However, aggressive handling can result in the degradation of proteins, aggregation, deactivation and sometimes even breakdown [9,70,71]. These are important mechanisms to know about and are covered within the next sections of the review.
Shear is present in most bioprocess unit operations . It has been published that these harsh hydrodynamic forces can change biomaterials [70,71]. Most fermentations encounter varying levels of hydrodynamic and shear forces due to the high dissolved oxygen demands required by growing microbial or mammalian cell suspensions [72-74]. Over time, mixing of the process broth in fermenters carried out by the impellers can cause these shear forces to disrupt biomaterials . The shear forces generated in these vessels can degrade the product, for example they break up soya protein precipitates . However, soya protein precipitates are large and it is assumed that large aggregates break up readily because they are very susceptible to micro-scale shear turbulence. Nevertheless, it has been shown that shear forces have detrimental effects on the stability of proteins and enzymes [9,71,77,78].
The complex structure of biologics, or biopharmaceuticals, compared to their smaller chemical counterparts may render them susceptible to process-related stresses; and hence low yields can result from shear related damage during bioprocessing [79,80]. The downstream processing should aim to separate the biological product from the bulk impurities cost effectively, generating high protein yields with high activity. However, protein functionality can be easily compromised by small structural alterations resulting from harsh bioprocessing conditions . The purification train used for proteins can vary, hence the level of hydrodynamic forces inflicted upon different protein products may produce varying levels of degradation. It is assumed that shear related loss can be particularity prevalent in centrifugation, types of filtration, chromatography, formulation and even the pumping of solutions [81-84].
Further shear losses may occur near the end of the purification train when the process stream becomes more purified and concentrated with product. This is because possible impurities that could potentially protect the proteins from loss have been removed earlier on in purification. Stabilising agents can be used in formulation strategies to reduce degradation and ensure product stability .
After fermentation an easy, fast and cost-effective way of fractionating the process broth is by the use of the disc-stack centrifuge. However, these industrial machines generate a large amount of fines compared to laboratory scale machines [10,82,86]. As a result, the ability to predict the clarification of an industrial disc-stack centrifuge from laboratory scale can be overestimated. This inaccuracy is thought to originate from the fact that industrial machines produce more shear than smaller laboratory centrifuges, therefore resulting in the production of more fines.
Membrane separation processes are used extensively in downstream processing as they can be applied to a range of tasks including filtration, concentration and buffer exchange. However high flow rates and turbulent air/liquid interfaces occur frequently during the diafiltration and concentration of protein solutions in ultrafiltration rigs. This continuous recycling and vibration action produce high shear forces that can result in enzyme inactivation [81,87].
Work by Bowen and Gan  showed that losses in enzyme activity occurred when the protein was under prolonged contact with the membrane during recycling under very low applied pressure. The results suggested that membrane/enzyme interaction generated shear-induced deformation of the enzyme structure with subsequent loss of activity . Where gravity cannot be used to move the process solution to another unit operation pumping must be used, which can generate unwanted shear. It has been shown that some pumps, like positive displacement and centrifugal designs, can inflict excessively high shear forces on protein precipitates and cause their break up . The literature suggests that the use of low shear pumping techniques like peristaltic and diaphragm pumps would reduce any disruptive effects on biomolecules [83,88]. Other work on shear in pumps by Virkar et al.,  suggests that secondary shear effects may cause protein degradation. These secondary shear effects may occur as air/liquid and/or solid/liquid interfaces, which are common during the bioprocessing of proteins .
Various authors have shown that different cells types like animal and bacterial are shear sensitive [6,53,89]. Results have shown that microbial suspension broths like bacteria and yeast, are significantly more robust during fermentation from shear induced degradation than mammalian cells. It is thought that this extra strength comes from the cell walls these microbes have [89,90]. Shear and air/liquid interfaces are commonplace in the fermenter and hence their effects could lead to protein degradation as seen in small scale studies [91-93].
In conclusion, there are many instances where shear and excessive high forces can be generated during downstream processing. It is the biochemical engineer’s responsibility to assess the best route for maximum protein production with highest yield and activity.
Harsh hydrodynamic effects and shear conditions can result in hydrophobic regions of proteins being exposed. Interactions between exposed hydrophobic regions may result in protein aggregation, which can markedly reduce the yields of biopharmaceuticals and presents a significant problem when working at high protein concentrations [94,95]. Insulin is a good example of one of the therapeutic products that suffers from aggregation. Work by Sluzky et al.,  investigated the causes of this aggregation during induced agitation of the protein solution. They found that the addition of a mixture of sugar and a non-ionic detergent reduced the level of aggregation. This is a very important idea as the addition of reagents to other process solutions could reduce the loss of product during downstream purification techniques .
It is a common perception in the literature that a shear field can somehow distort the enzyme/protein molecule permanently, changing its native structure [70,97-99]. Therefore, deactivation can be defined as the irreversible effect a shear force might have on a protein, resulting in the loss of functional activity or native structure . A variety of proteins and enzymes have been shown to lose activity under the effect of shear [70,71,77, 87,100,101]. This form of product loss may occur often when purifying protein solutions at production scale because large volumes of very concentrated solutions are often moved around the plant. Another mechanism of deactivation could be that shear does not denature protein, but only slightly distorts the shape of the enzyme, changing the shape of the active site and therefore altering the enzyme kinetics of the protein.
Temperature is also an important factor in protein deactivation. Lencki et al.,  shows that increasing the temperature from 25 to 30°C increased the shear deactivation of dextransucrase. In the same study they looked to alter proteins, carboxypeptidase and chymosin, and their results showed that dextransucrase was much less susceptible to shear effects compared to the other two proteins. This shows that different protein shapes and processing conditions could lead to varying levels of deactivation. This variation in protein stability further justifies the reasons for the present line of work, as detailed analysis of the MFECP1 fusion protein will hopefully uncover ways to reduce deactivation.
Several publications have shown that high shear effects can breakdown biomolecules such as proteins and plasmids [9,68,80,96,102,103]. Maa and Hsu  showed that two proteins, Recombinant Human Growth Hormone (rhGH) and Recombinant Human Deoxyribonuclease (rhDNase), were shear sensitive in different ways. The rhGH generated different thermal properties after shearing which suggests that this protein had undergone shear induced conformational change. Also, SDS-PAGE gel analysis of the same protein showed that shear caused peptide bond breakage. Interestingly, none of these changes were found in rhDNase, confirming that different proteins behave differently under shear conditions. The variation in the results between the two proteins must have been to do with the structural differences, as rhDNase contained sugars within the protein structure, whereas rhGH does not. Overall the results confirm the earlier hypothesis that protein stability is directly related to the physical structure of the protein.
Further work by Levy et al.,  using a Perspex rotating disc shear device 40 mm in diameter, showed that large biomolecules like plasmids were prone to shear damage. The shear conditions that were chosen occurred during the processing of plasmid-based gene therapeutics and DNA vaccine production. The findings showed that the tertiary structure of the plasmid can be severely affected by shear forces. The extent of damage was dependent upon the size of the plasmid and the ionic strength of the environment . The results of these studies show that detailed characterisation of protein solutions during shear conditions are required . This is the only way one can deduce if proteins degrade and whether this type of damage can be reduced.
So far during this literature review, considerable emphasis has been directed on how biopharmaceuticals may degrade during their production and purification. However, the aim of this review is to understand any mechanisms that degrade the proteins using USD technology , figure 9 shows an Ultra scale down shear device. This knowledge may then be used to reduce similar product loss at a larger scale . It is estimated that it costs £0.5 billion to bring a new drug to the market and the process takes nearly 10 years . To retrieve this investment in a relative short period of exclusivity it is of value to get through drug manufacture as soon as possible [16,108]. Some of the approaches used by major companies are to invest heavily in pilot scale capacity, which is costly, time consuming and labour intensive. The USD work hopes to reduce some of these burdens, thereby reducing the price of these drugs, allowing more people to be treated faster. This can be achieved by developing on the methodologies used previously in other studies [11,12,86,109,110] to show that ultra-scale-down technology has the potential to be used as a rapid tool for bioprocess optimisation, reducing development times and costs .
Figure 9: Shows the Ultra-Scale-Down (USD) shear device that can be used to shear all MFECP1 fusion protein samples. The Euro shows the scale of the device, a tripod stand was used to stand the device in an ice cooled water bath. Electrical cables from the top allow the 7.2 V 500BB race VS motor (Graupner, Germany) to rotate the disc in the chamber. The total volume of protein solution occupied was 20 mL.
The ultimate goal of scale-down technology is to simulate the working conditions of large-scale processes and to predict outcomes on scale up. This involves a detailed understanding of the stresses the large-scale unit operations may inflict on the fragile biomaterial and also knowledge of the characteristics of the protein . Scale down knowledge is particularly helpful at early stages of drug development, for example from phase 1, where only small quantities of test material may be available. As the trials progress, larger quantities of process materials are required, because the numbers of patients in each trial increases. If a robust and reliable scale-down methodology was produced it could reduce process development times with significant cost savings .
All these possible financial problems fuel the incentives for better scale-down technology, further justifying the present line of work, which will try and mimic the harsh hydrodynamic conditions a MFECP1 fusion protein would encounter during bioproduction.
Research to date
The scaled down fermentation process has been well described [112-114]. However, there is less detailed information published for downstream unit operations because specific purification trains vary depending on the product characteristics. Straight forward scale-down of a fermentation unit is accomplished by geometrically maintaining the size of the tank and impeller ratios and Camp number between scales of operation ; this works particularly well for protein precipitation . Other examples of unit operations that have been scaled down include the very first scale-down of an industrial disc stack centrifuge, where blocking of discs decreased the centrifuge capacity by 10-fold [116,117]. Although this work successfully demonstrated the scale-down of a large-scale centrifuge, the method still required litres of process solution. A more convenient technique would operate at bench scale to ascertain important large-scale production parameters, for example using millilitre quantities as shown by .
A variety of other scale-down accomplishments have already been identified in the past that have been shown to predict large scale unit operation. These include centrifugation, EBA capture, and filtration membranes [11,12,109,118,119].
It is thought that in the future, a series of scale-down mimics of downstream unit operations could be linked together to create a whole scale down bioprocess train. This, coupled with small scale fermentation, could demonstrate pilot scale working parameters using very small volumes of process material early on in drug manufacture . This last piece of work is set to be the future of scale-down as it uses much smaller quantities of process fluid. It is sometimes referred to as automated microscale processing and is reviewed in the next section.
The ultimate goal of scale-down models and devices would be to harness them in multi-tasked robots that have the ability to analyse numerous sample volumes (at 96 well level), with varying operating conditions. This rapidly emerging area of research is termed microscale processing and, just like ultra-scale-down, it increases the speed of bioprocess design by reducing the amount of volumes required [106,107]. It is likely that whole bioprocess sequences could be operated in microwell format, allowing high throughput automation analysis of very small quantities of therapeutic entities. Optimum candidates that are better suited for large scale production can be carried forward, whereas prospective products that have high potential for failing due to their shear sensitive nature can be removed. This could save a pharmaceutical company considerably millions, in avoiding expenditure on drug candidates that do not pass late clinical phases due to scale up issues [16,108].
If this technique is successful it would result in generating drug candidates faster, thereby increasing the duration of patents which would produce immense financial success to the industry. This shows how future scale-down models could have great value in the biopharmaceutical industry. However, at the moment there are many problems with going down to the microscale level, sampling, oxygen and mass transfer to name just a few [106,107]. These challenges can be improved by conducting further research into microscale technology.
Conclusions of ultra-scale-down technology
The reasons for the improvement of ultra-scale-down technology is to have a better understanding of how to produce complex shear-sensitive biopharmaceuticals early on at large scale during the drug development chain [11,106]. Ultra-scale-down technology incorporates the ability to use very small quantities of sometimes expensive drug product to identify desired operating conditions. At present several scientists have already used scale-down research to identify optimum large-scale operating conditions, for example in the use of centrifuges, specifically the feed inlet , feed outlets  and clarification . It is hoped in the future that ultra-scale-down techniques can be used to investigate how robust antibody fusion protein are and show how this knowledge can be used to try and improve its large-scale production . Dilute suspensions of the test biomaterial could be characterised in the shear device to identify conditions that degrade the product. Next, this forced degradation could be reduced through the addition of different reagents, proteins and/or detergents . The most promising reagents could be taken forward to the large scale and possible areas of yield improvements could be explored.
The present worldwide publication is wholly dedicated in honour of my late mother and father Baljinder Kaur Blas and Ram Blas who both passed away during the course of my PhD. It is also dedicated to my beautiful children, Leah Sophie and Theodore Colin Blas who inspire me to achieve the very best in life. I would also like to thank my fiancée Miss Tiffany Amelia Greenwood, your unconditional support means the world to me.
Furthermore I would like to thank the following scientists for their trust, guidance and the opportunity to complete this study:
Professor Gary Lye (Head of Department, Biochemical Engineering at UCL),
Professor Nigel Titchener-Hooker (Dean of Faculty of Engineering Sciences at UCL)
Professor Kerry Chester (Research Department of Oncology, Cancer Institute, UCL)
Professor John Ward (Synthetic Biology for Bioprocessing, UCL)
Professor John Mitchell (Communications Systems Engineering, Vice Dean Education, UCL)
Professor Nik Willoughby of Bioprocessing, Heriot-Watt University
Professor Mike Hoare, UCL.
Copyright © 2019 - All Rights Reserved - ocimumpublishers.com