GEN Exclusives

More »

Feature Articles

More »
Aug 1, 2007 (Vol. 27, No. 14)

Strategies for Ensuring Optimal Scale-up

Little or No Room for Error When Profitability Hangs in the Balance

  • Scale-up has always puzzled nonengineers, for whom terms like heat transfer and mass transfer engender a glazed-over look. With the added dimension of living production systems, what is true for scaling up production of foods, chemicals, and small molecule drugs is at least doubly true for bioprocessing. Bioprocessors scaling up are concerned not just with multiplying volumes of cells but in maintaining those cells’ productivity over extended periods. Nothing short of profitability hangs in the balance.

    “It’s all about speed to the clinic,” says Hank Talbot, Ph.D., who directs Dowpharma’s ( Pfenex Expression Technology™ microbial fermentation. Dr. Talbot avoids the catch-phrase “platform technology” relished by cell culture experts but that is what the Pfenex service is after all. In the cell culture world, bioprocessors prefer standard CHO cells, fermentation protocols, and downstream processing for all products within a particular molecular class, for example mAbs.

    Pfenex attempts to do the same for non-glycosylated proteins and protein fragments. Using Pfenex, Dowpharma can build a production strain from scratch in a few weeks, Dr. Talbot reports. Development begins with an array of 80-plus unique host strains of Pseudomonas fluorescens, crossed with different expression vectors and secretion leaders.

    This stage is carried out in microtiter plates. Candidate strains generating the highest yield are optimized at the 300 to 1,000-mL fermentation scale. “We can have a scalable process confirmed in 20-liter fermentors within a few weeks,” Dr. Talbot says. Dow does not perform manufacturing, but hands off the fully scalable process to contract manufacturers or customers.

    A key component of scale-up for many protein targets in Pfenex is optimizing the secretion leader. Dowpharma works primarily with secretion leader molecules containing disulfide bonds, which are best formed in the periplasm, the region between the bacterium’s inner and outer membranes. A secretion leader is a peptide that assists in protein transport across the inner membrane. As it passes through, peptidases cleave off the leader, leaving high concentrations of properly assembled, folded product in the periplasm. From here, the product is released through chemical and physical means developed by Dowpharma without disrupting the cell. This minimizes the release of host cell DNA, protein, and cytoplasm contents, which greatly simplifies purification.

    Pfenex fermentations use defined culture media free of antibiotics, animal-derived components, and organic nitrogen. Because it contains only ammonia, salts, and a carbon source, the process is readily scaled regardless of the protein product. Tweaks for improving soluble protein expression and/or isolation are worked out at small scale and directly applied during scale-up.

    Microbial systems are innately simpler than cell culture, which greatly simplifies technology transfer. Fermentations run about two days, compared with two weeks for mammalian cell culture. Secretion leaders help produce much cleaner feed-streams compared with nonsecreted microbial products, and volumetric productivity can be as high as 15 grams of protein per liter.

    Since it utilizes bacteria, Pfenex cannot produce glycosylated proteins. However, it can produce mAb fragments containing the active regions of these proteins. These proteins can be conjugated with toxins or other bioactive molecules or PEGylated to improve pharmacokinetic properties.

  • Critical Support Components

    Scale-up efforts that focus strictly on unit operations and engineering algorithms may still miss critical support activities, without which processes cannot move forward.

    Analysis, including process analytic technology (PAT) and more traditional QA/QC and process monitoring, play an integral role in scale-up. The larger the process the greater the investment and hence more tightly processors must control product characteristics. “We see more delays and constraints from product comparability characterization than for process-related events,” says Friedrich Nachtmann, Ph.D., head of biotech cooperations at Sandoz (

    Sandoz maintains a broad portfolio of products that it manufactures for customers as well as its own medicines. Most are produced through microbial fermentation, although the company also maintains a cell culture facility for mAbs. “There is more excitement with microbials, however, because product variability is higher in that production system than with cultured cells,” Dr. Nachtmann adds.

    To eliminate the risk of product quality drifting from batch to batch or during scale-up, Sandoz is investing significantly in “up-front analytical characterization,” according to Dr. Nachtmann. Most of this work is performed offline in parallel with process development. Sandoz also has a PAT program in place for achieving continuous process improvement of established processes. However, Dr. Nachtmann distinguishes PAT-related monitoring from real characterization of product purity.

    Data is another easy-to-overlook issue in scale-up. According to Pete Latham, president of Biopharm Services (, pharmaceutical/ biotech manufacturers spend 13% of their time looking for data and fail to find it between 10% and 20% of the time. Data problems plague the pharmaceutical value chain from discovery through manufacturing, where lost data means duplication of effort. Biopharm Services recently received a grant from the U.K. government to study how data is managed and used during process development and scale-up.

    Numerous electronic data-management tools already exist for handling manufacturing operations. However, current practice is not as unified as one would imagine from reading data-management vendor literature. In the real world, end-users rely on electronic laboratory notebooks, off-the-shelf data programs like Excel and word processors, and old-fashioned paper records. “This creates islands of data,” says Latham. “In fact there is no comprehensive system for managing and leveraging that data consistently enough to serve the tech transfer needs of bioprocess scale-up.”

    Biopharm Services is working on a new data architecture wrapped around ISA-88, the international standard for flexibility in production, which Latham hopes will make manufacturing data more amenable to scale-up situations. “ISA-88 is a methodology for collecting and putting data into a format that is scale independent.” The standard employs a modular design approach that speeds and facilitates building process definitions and reuse of previously defined processes and scale-up logic.

    From this data, one may apply transition algorithms that translate process details to fit any type of equipment or facility. Ultimately, the goal is data-management software that facilitates scale-up and technology transfer. “One way to look at this is to take general process data and translate it into a master recipe,” adds Latham.

    Don’t software packages already do this? “Yes, ERP software and lab notebooks, in particular, facilitate scale-up and technology transfer, but users lack sufficient faith in ELNs, which means that most people keep paper records side-by-side with electronic records. Biomanufacturers desperately need an impetus to abandon the existing paradigm,” concludes Latham.

    The other problem with a universal data system is the unique value system for pharmaceutical and biotech drugs. A one-size-fits-all mentality simply doesn’t apply to bioprocess scale-up, nor are suited to “hard” process industries like semiconductors particularly amenable to biotech. “We can’t just put the process into a bigger pot and get the same result,” adds Latham.

    For these recurring questions of scale, Biopharm Services BPS is involved with the U.K.’s Innovative Manufacturing Research Centre for Bioprocessing at University College, London, which has been building scale-down models for some time.

  • Troubleshooting

    Scale-down is also used for trouble-shooting problems in large bioprocessors. Last year Invitrogen’s ( CRO division in Grand Island, NY, began using a SimCell®, a scale-down microbioreactor from Bioprocessors ( to optimize cell culture media from development through full-scale processing.

    Increasingly, services like media optimization are carried out by contract research organizations including Invitrogen. Mammalian cell culture media contains up to 100 components. Testing each ingredient, or even the dozen or so critical ones, for each cell line and product, is a major project in and of itself. Even microbial fermentation media, which may hold up to ten components, requires systematic optimization. Bioprocessors must maximize yield as well as keep track of the consistency and source of raw materials.

    SimCell is a high-throughput, simulated scale-down bioreactor tool that runs a large number of cell culture experiments in parallel. SimCell provides multiple micro-bioreactors ranging in size from 150 microliters to 1.5 mL, allowing researchers to test multiple cell culture conditions simultaneously, according to the company.

    Invitrogen uses the Bioprocessors device to provide services related to development and scale-up of cell culture media. “SimCell gives us tighter control over the process than is possible using spinner flasks,” says Bob Burrier, Ph.D., vp for R&D at Invitrogen.

    Cell culture media has been described as the single most important factor in the rise in mammalian cell culture productivity. “The majority of innovation going forward with higher titers will occur in media development,” reports Dr. Burrier.

    As a leading media provider, Invitrogen has a vested interest in retaining customers from process development through manufacturing. Nearly all its large-scale media products are custom-designed, and increasingly these products are delivered as powders and reconstituted before use, according to Dr. Burrier. Many Invitrogen media products are chemically defined, but some still contain “undefined components.”

    Through SimCell, Invitrogen can do clone screening and then optimize a process for pH, agitation rate, dissolved oxygen, and feed strategies. The company is planning to add glucose utilization in the near future. “One of the nice things about SimCell is media and process conditions can be tested at a specific cell density and users can watch what happens in a high-throughput way,” Dr. Burrier notes.

    Merck ( uses scale-down-suitable bioreactors from a variety of vendors. According to Beth Junker, Ph.D., senior director for fermentation operations, the company applies scaledown methods for both new and existing processes.For the former the goal is process optimization, while for the latter the objective is continuous process improvement. Process analytics play a role in this exercise as well. “We look for sources of high variability that impact product quality and seek to reduce them,” says Dr. Junker.

    With respect to scaling down an entire process, the challenge is to devise scaledown experiments that accurately represent large-scale processes and that, at some level, can predict in a bioreactor of a few liters what will occur at thousands of liters.

    Another issue is how to apply engineering equations to scale-down, particularly what the small-scale conditions should be when operating in trouble-shooting mode and which large-scale parameters apply when trying to predict from scale-down to full-scale runs.

    “Operating at smaller scale permits greater throughput and experimentation,” Dr. Junker adds, “while minimizing experiments at large scale.”


GEN Jobs powered by connects you directly to employers in pharma, biotech, and the life sciences. View 40 to 50 fresh job postings daily or search for employment opportunities including those in R&D, clinical research, QA/QC, biomanufacturing, and regulatory affairs.
More »

Be sure to take the GEN Poll

Global Antibiotic Resistance

Do you find the association of a country’s antibiotic resistance being tied to its degree of corruption rather than its wealth surprising?

More »