Adapting Industry Practices for the Rapid, Large-Scale Manufacture of Pharmaceutical Proteins

Author: David Estell

The tools and methods of industrial biotechnology could be adapted to produce pharmaceutical proteins for a pandemic influenza vaccine.

At Genencor, an industrial biotechnology company, full-scale industrial manufacturing of proteins at volumes of 30,000,000 gram-active proteins per month is possible within weeks of the creation of a final protein-based molecule, and prices are less than $1 per gram-active protein. By comparison, pharmaceutical production volumes are routinely less than 10,000 gram-active proteins per month, selling prices are higher than $1,000 per gram-active protein, and full-scale manufacturing timelines are more than a year from creation of a final protein-based molecule.

An important aspect of industrial protein manufacturing is the development of the production process before the creation of the final product molecule. This is essentially the opposite of the classical pharmaceutical approach in which the product is created, and the manufacturing process is then developed from the research chemistry bench. In industrial protein manufacturing, an initial robust process is developed for a scaffold protein that has many of the desired properties; then a few changes are made in this protein to create the final molecule. The production and purification processes developed for the scaffold protein can then be modified for the final molecule. Industrial protein manufacturing can supply a protein-based product at yields, volumes, and cost levels not possible with the classical pharmaceutical approach.

The tools and methods of industrial biotechnology could be adapted to produce several million doses of pharmaceutical proteins for a pandemic influenza vaccine, all within a fairly rigid time frame for an effective vaccination campaign. The technology is available, and industrial costs and volumes would provide incentives. Industrial production costs and volumes could also enable alternative delivery systems that are less protein efficient, but have simpler logistics, such as topical, oral, and inhaled delivery systems. The prerequisites are in place for the pharmaceutical industry to take advantage of these incentives, and the time to act is now.

Protein Production
Protein engineering begins with the identification and assembly of a protein scaffold. In the case of influenza vaccines, protein scaffolds (the influenza proteins that contain the epitopes) are already identified, because influenza strain selections are made and formulation cultures prepared from the Center for Biologics Evaluation and Research/Centers for Disease Control and Prevention (CBER/CDC) reassortant “seed lots.” Every year, the scaffold is modified as regulatory agencies add different epitopes for the strains that have changed. Insofar as the production process is already in place for the scaffold proteins, which are modified to have new epitope sequences every year, the production of influenza vaccines already follows the general principal of industrial protein production.

It is important that the gene/host system be able to produce a desired protein in abundance with a minimum of unwanted by-products. It is also necessary that the high-yield fermentation process maximize end-product output from the gene/host system. From the same protein scaffold and basic protein, a robust, rapid, and efficient recovery process can then be developed. The fermentation and recovery processes are designed to be completed within hours or days to achieve the high throughput necessary for industrial manufacturing.

Formulation and delivery systems must allow flexibility and end-product stability. Protein-based end products that are transported and stored around the world must be able to withstand the high temperatures and humidity of tropical locations without loosing potency, strength, or efficacy and must remain stable for years with no refrigeration. Finally, the protein scaffold must be engineered to provide the desired properties. In the case of the influenza vaccine, epitopes for the new strain are added to the protein scaffold, which is then introduced to the protein production pathway.

Protein Scaffolds
The protein scaffolds listed here have been created and engineered by Genencor; for each scaffold, a high-yield fermentation process and recovery or purification process has been developed. Multi-domain fusion proteins, monoclonal antibodies, viral or bacterial coat proteins, protein-based inhibitors, and enzymes have been used in an array of industrial formulations, many expanding the application and use of genetically engineered processes to new domains as varied as treating textiles, processing milk to make cheese or yogurt, cleaning contact lenses, processing paper, brewing low-calorie beer, boosting the nutritional value of food, and converting plants such as corn to chemicals such as ethanol.

Multi-Domain Fusion-Protein Scaffolds
It is most desirable to express the protein of interest properly folded, with no refolding, and in a clean medium. A properly folded protein structure has a tertiary or quaternary three-dimensional shape in a functional, stable conformation (Figure 1- see PDF version for figures). In general, proteins expressed using E. coli as the host production organism do not express the properly folded protein with great efficiency. Other bacterial/fungal systems are better equipped to act as the host production organisms.

The desired foreign protein can be produced/secreted at very high levels using the host’s own expression system simply by linking the foreign protein to a protein already secreted in abundance by the host organism. The “fused” protein goes through the production/secretion machinery of the host to produce properly folded product protein in very high yields.

Aspergillus niger, one of the most commonly used host production organisms, secretes very high levels of glucoamylases. Yields of 30 grams per liter, at rather fast fermentation rates and at very low production costs, makes this an extremely efficient microbial expression process (Figure 2). Glucoamylase, the fusion protein partner for foreign protein (e.g., vaccine or antibody) production, is an enzyme (a protein that contains amino acids), that is a multi-domain protein. Glucoamylase has a starch-binding domain separated from the catalytic domain by a linker. The glucoamylase pulls the starch-binding domain through the secretion machinery. The foreign protein is then expressed in the properly folded structure, regardless of the complexity of the protein. Thus, this is the system of choice for the expression of any antibody or vaccine candidate.

Gene/Host Systems
A multitude of systems have been used for large-scale, efficient expression of proteins or enzymes. The gene/host production organisms listed here all secrete fused proteins through the secretion machinery of the host to provide properly folded, secreted, product proteins in high-level yields. Bacterial examples include, Bacillus subtilis, Bacillus licheniformis, and Bacillus lentis. Fungal host expression systems include Aspergillus niger and Trichoderma reesei.
Each gene/host production organism has been characterized, most of the genomes sequenced, and the expression systems modified to enable the system to express non-native genes with high efficiency. Knockout strains have been developed and metabolic pathways altered so that the host utilizes simple carbon sources in manufacturing the desired protein in a process called metabolic pathway engineering. The fermentation processes are rapid, and the downstream processing maximized and efficient. These are robust, commercially viable formulations that are easy to scale up and transfer to a customer, and costs are competitive.

Microbial/Fungal System
One advantage of microbial or fungal systems is the speed at which stable host production strains can be constructed. Instead of six months, it takes less than two weeks for bacterial systems, and approximately four weeks for fungal systems, to prepare a seed bank and make it available for high-yield fermentation processes. Another advantage is that screening for improvements in the manufacturing process can be accomplished using the host organism. This translates into higher efficiency and higher yield. Fermentation times are shorter than for cell-culture systems, the processes are robust, and the capital expenditures are lower than for cell-culture systems. As a result, the cost of goods sold is also lower.

Fermentation Systems
In bacterial systems, the typical fermentation time or process turnaround time—from the time the seed is started until the time the production fermenter is removed from operation—is three days; the tank is ready to be placed into operation again on the third day (Figure 3). In fungal systems, the turnaround times can be as long as 20 days, with an average of 10 days. Production fermenters generally hold 30,000 liters to 300,000 liters; a few recovery pilot plants use 3,000 liter fermenters for smaller volume products.

Protein-Recovery Processes
The recovery or purification process accounts for a large portion of pharmaceutical production expenses. In industrial conditions, expenses can be limited, as long as clean media are used during production recovery, and as long as the processes have been maximized to produce high yields. In industrial applications, large-scale filtration and extraction are used to recover bulk protein. Chromatography, which can also be used, is scalable to recover the bulk of large-molecule proteins. As in some small-molecule applications, crystallization can be used to purify the protein. Initial purification by crystallization, which is also scalable, often yields a product of greater than 95 percent purity. All of the recovery systems already being used for injectable antibiotics produced through fungal fermentation can also be used for proteins produced at the same yields in industrial production.

Formulation and Delivery Systems
The technology is available to provide stable products in delivery systems that can withstand extreme conditions (Figure 4). In the past, entire shipments of frozen pharmaceutical products were lost during shipping delays or power outages, not to mention the lack of storage facilities in remote locations. Genencor integrates final formulation of the protein as part of the process-development strategy, so protein formulations can be shipped around the globe under extreme conditions. Temperatures during shipment can be anywhere from below freezing to higher than 40?C, but the performance and stability of the active ingredients remains efficacious for the life of the product, including during shipment and storage, in either solid or liquid form. These protein formulations do not require refrigeration or freezing.
Unique properties of solid, multilayer, granular formulations can control the release of active ingredients and delay degradation by humidity, temperature, and other environmental influences. Compartmentalizing the active ingredient against other components of the product may also help maintain its effectiveness and control the release of active ingredients. An example everyone knows is the “tiny little time pills” in Contac? cold formulations.

Formulation and delivery systems are currently engineered to meet strict quality and regulatory requirements for food-grade products. The specific needs of the customer dictate the level of stability and sophistication of the final formulation.

Typical Microbial Production Process
The production of a vaccine by typical microbial processes requires two to four weeks to create the production host (two weeks for a bacterial system and up to four weeks for a fungal system), at which point the seed bank can be created. The fermentation process can take anywhere from three days in a bacterial system to 20 days in a fungal system. Recovery and formulation can take from two to ten days. This is a routine timetable using 30,000 to 300,000 liter vessels and producing tremendous amounts of protein.

Rapid Protein Drug Production
The first step in applying industrial techniques to the production of a vaccine is to ensure that a gene/host expression system is in place for the protein scaffold; the base scaffold should be part of the drug. For an influenza vaccine, the scaffold should be the epitope containing proteins from the influenza virus strains for that year, engineered so that the basic properties are in place. The engineering of proteins for the desired properties can also include a predictive assay for immunogenicity (Stickler et al., 2000) and a process by which the immunogenicity can be increased or decreased for the desired protein. Other properties include required pharmacokinetics. All of these properties are engineered into the basic protein scaffold. Next, a high-yield fermentation process is developed for the scaffold. Finally, formulation and delivery processes are engineered.

When a request is made for several million doses of a product with certain properties, the designed molecule can be linked to the multi-domain protein and produced with the pre-existing high-yield fermentation system. In a protein drug, for example, only a few changes in the sequence of a protein scaffold would be necessary to make a revised protein drug to meet a new need. For example, if there is an existing antibody, then only the variable regions (i.e., the binding sites of the antibody on the antibody scaffold) would have to be changed to make a new antibody that can recognize the very specific new antigen. If a new vaccine must be made to counter a pathogen with different peptide sequences (antigenic drift), the new sequences can be added in much the same way as they are in current flu vaccine development; then the epitopes (three-dimensional surface features of the antigenic molecule) can be linked to the same scaffold and introduced to the fermentation system.

The same is true of a new drug with new target binding sites to act as inhibitors in a metabolic or chemical reaction. The new target binding site or receptor binding site is engineered into the protein scaffold and introduced to the fermentation system, which is already in place for the protein scaffold.

These are all variations of the same biotechnology platform, that is, protein engineering of “biomachines” to manufacture desired product proteins or enzymes and formulated to remain stable and effective over the shelf life of the product. Manufacturing platforms could be established now for known pathogens and licensed pharmaceutical drug products. Regulatory groups could review and approve gene/host systems, fermentation and recovery processes, as well as analytical testing processes and release criteria. If these steps were completed now for pharmaceutical drug products already established in the market, when new drug products are identified (e.g., new viruses, new pathogens, new target binding sites), the process could be modified and the regulatory filings revised. This process would be similar to the current yearly process for influenza vaccine production. The FDA would then be reviewing familiar processes, and approval time would probably be shorter.

Glycosylated Protein Production
Glycosylation is the process by which polysaccharides (complex carbohydrates, such as starch, glycogen, and cellulose) are added to proteins. Most protein drugs on the market today do not depend on glycosylation for their activity. Currently, proteins for which glycosylation is not required for efficacy of the protein can be produced by existing engineered pathways. The preferred way to produce these proteins would be as non-glycosylated versions. However, proteins for which glycosylation is important may require additional gene/host engineering or post-production modification.

If a rare protein requires an absolute replication of the sugar sequence in order to be efficacious, either the host would have to be engineered to duplicate the exact sequence of polysaccharides, or the polysaccharide unit would have to be added post translationally, which presents a separate problem.

Dose-Related Protein Drug Production
Vaccines and other products with individual dose sizes of <100 mg can be made with existing technology and capacity; yields of 1 gram per liter are all that is required. Products that require dose sizes of >100 mg (e.g., monoclonal antibodies), which are rare in the pharmaceutical industry, may require initial improvements in the host system for the scaffold protein to ensure that yields are high enough to provide enough protein for several million doses of vaccine.

Current industrial biotechnology production processes are capable of producing 100,000,000 grams of protein in less than 12 weeks. However, to appropriate the use of biotechnology capacities, all aspects of protein manufacturing processes must be in place, including protein scaffolds for the gene/host systems, high-yield fermentation systems from which robust recoveries can be made, and final formulations that allow transportation and storage under extreme conditions. Yields must be at least 1 gram per liter to meet timelines. The industry has the capacity and the incentives to do this.

The same is true for capacities in the injectable and oral antibiotics industry, which currently manufactures product through fungal fermentation. This industry works closely with the FDA to meet regulatory requirements and improve antibiotic processes continually to reduce the necessary capacity. This is a $13 billion per year industry with sales averaging $1 to $100 per gram product.

The antibiotic fermenter capacities available in this industry or the biotech industry would more than meet the needs for manufacturing proteins by the pathways discussed in this article. The fermentation processes for production of antibiotics or industrial proteins and enzymes are comparable to those anticipated for protein production pathways. The same is true for recovery and purification processes. The tools are available, and the time is now.

GEN (Genencor International). 2006. Formulation Delivery Systems. Available online at:
Stickler, M.M., D.A. Estell, and F.A. Harding. 2000. CD4+ T-cell epitope determination using unexposed human donor peripheral blood mononuclear cells. Journal of Immunotherapy 23: 654–660. Available online at:

About the Author: David Estell is vice president of technology for Genencor International.