August 1, 2006 (Vol. 26, No. 14)
Susan Aldridge, Ph.D.
Improvements in Titer Shift the Bottlenecks Downstream to Purification
Mammalian cell products dominate the biopharmaceutical industry and will continue to do so, predicted Andrew Sinclair, managing director of BioPharm Services (www.biopharmservices.com), at this year’s “bioLOGIC” conference, held recently in Amsterdam. The top-four product types in clinical trials are monoclonals, cytokines, recombinant proteins, and increasingly, vaccines.
Input into clinical trials is driving growththere has been an 8% increase in INDs between 1988 and 2002. Now there are two million liters of manufacturing capacitya four to five times expansion on 2001, which is roughly in line with requirements.
Improvements in titer mean that the focus now is not so much on capacity as on purification. The big challenges now lie in downstream processing. “There are lots of ideas out there,” Sinclair said. In both downstream processing and in other stages of manufacture, scale up can present a challengebut one that can be met by employing the right technologies.
Scale up generally involves taking a lab-scale procedure and replicating it as closely as possible to get larger amounts of product as specified either by a client or the regulatory authoritiestypically going from a one-liter shake flask to 50 liters and then to 500 liters in a fermenter.
The use of chromatography in purification can lead to some challenges in scale up, which may be solved by new approaches. Sylvio Bengio, Ph.D., scientific communications manager at Pall (www.pall.com), described how the mixed-mode Hypercel media can exploit different mechanisms of interaction with a target protein. The MEP, PPA, and HEA ligands attached to the media bind by hydrophobic interaction (MEP also binds by affinity) and desorption is driven by electrostatic charge repulsion. The proteins on the column are separated by both isoelectric point and by hydrophobicity, allowing a purification to be fine-tuned.
“This approach can provide new solutions, for instance, if the conductivity of the feed is too high for traditional ion exchange, when the affinity ligand is too expensive, or if the process would otherwise need too much salt,” explained Dr. Bengio. MEP Hypercel is an alternative to Protein A for antibody capture, eliminating endotoxin and residual DNA, he noted.
The Hypercel media have also been applied to recombinant proteins and enzymes and have proven valuable in separating aggregates from monomer. One application has been in the purification of the antistaphylococcus protein recombinant lysostaphin for BioSynexus where the protein, needed for Phase I and Phase II, was captured and cleared of endotoxin in a replacement of an ion-exchange step.
“Mixed mode is a new tool for process development that is based upon a scalable matrix. Compared to conventional chromatography it can allow binding of the target with less or even no salt,” said Dr. Bengio. “The ligands are robust and synthetic, so they can withstand harsh treatments.” Of course, using these new media requires, initially, more process development time than Protein A and ion exchange, but they offer an advantage when conventional media do not work as well at large scale.
Medium optimization is an important aspect of cell culture, especially when it comes to scale up. Six sigma design-of-experiment methods were used to discover how medium and feed components interact with bioprocess parameters but their full exploitation has been limited, until now, by the tools available96-well plates, shaker flasks, and benchtop bioreactors.
The first two are cost-effective for this work but data from them is not predictive of scale up. Bioreactors are scalable by definition but are usually limited in number. Although bioreactors give better data, their use is costly and time-consuming. “We know from experience the experiments we would like to do for our customers,” said Reg Joseph, business area manager for Invitrogen’s (www.invitrogen.com) GIBCO bioproduction systems and service division.
However, to do all these experiments was, until now, cost-prohibitive. Joseph described new research on medium optimization, which the company has carried out with the SimCell, a microbioreactor array (MBA) developed by BioProcessors (www.bioprocessors.com). Engineered to simulate the conditions of stirred tank bioreactors, the MBA consists of six 0.6-mL chambers, each of which can contain an experiment (with different pH, medium, and so on, monitored for cell growth through optical density) that would otherwise be done in a benchtop bioreactor.
The GIBCO R&D department conducted an experiment with the system to determine variance between the different chambers and verify scalability of the effect of pH in a CHO cell line. The results showed that the outcome in a stirred tank bioreactor is accurately predicted by the SimCell.
Next, the system was tested in a complex experiment involving four different media components, salt, three feed components, and pH as the process parameter. This experiment showed the power of the system to detect interactions between components of the system, confirming the SimCell as a powerful new tool for medium optimization, according to the company.
Filtration is necessary at many stages of a manufacturing process, so an understanding of filterabilitythe filtration behavior of a solutionis crucial. Filterability testing can be used to determine the capacity of a filter system with the aim of determining the filter area that would be needed to process a batch without blocking the filter upon scale up.
Domnick Hunter (www.domnickhunter.com) offers filter cartridges and single-use capsules, among other products, as well as advice for scaling up a filtration process. The company’s approach involves, first, small area filter testing on discs of membrane to determine the filterability of a solution under either constant pressure, which is quick, or constant flow, which is more accurate, or both.
“This gives an initial idea of what might work,” said applications specialist Jon Houseman, who carries out on-site studies of filtration scale up for the company’s clients. Then the study is done on pleated products (capsules with a larger surface area) followed by full-scale cartridge trials.
In both stages, disc evaluations are carried out concurrently to make sure the solution is still behaving as in the initial test. Even small changes in factors such as pH can alter filterability, with buffers tending to be more predictable than media. “Filterability trials should be done at every stage of scale up,” advises Houseman. The structured approach to understanding filterability is the key to successful scale up.
Researchers at The German Research Center for Biotechnology, Braunschweig, and Miltenyi Biotec (www.miltenyi.com) have been looking at a new cell separation technology based upon the hydrocyclone, which can be used as a perfusion system in large-scale animal cell culture. The hydrocyclone works like an inverted fixed-wall centrifuge without a rotating shaft.
Until now, the application of the hydrocyclone has been to yeast but now the researchers have developed a smaller system for shear-sensitive animal cells, which has been tested on HeLa and SP2/0 cell lines in serum-containing media and on CHO and hybridoma cells in serum-free media. Scale up to 20- and 200-liter production was relatively simple with no adverse impact on cell viability.
Meanwhile, CMOs like Angel Biotechnology (www.angelbio.com) deal with a whole range of scale-up issues for clients. “Rapid development of scaled-up, GMP-compliant processes for the manufacturing of proteins is a key activity. Many of our clients require several grams of investigatory medicinal product (IMP) to support their clinical trial program, while much larger quantities of product are required for commercial sale,” said Nigel Shipston, head of GMP process development.
The experience and flexibility of companies such as Angel is important because small drug discovery companies usually don’t have the experience, time, or capital to establish a process for manufacturing sufficient quantities of their product to the GMP standards now required by EU Clinical Trials Directive 2001/20/EC.
Moreover, important manufacturing constraints can be overlooked, and this can have enormous impact in later stages. Clients often develop laboratory-scale strategies that can produce a few 100 mg of protein. They then realize these procedures are difficult to scale up in their own facility, even to complete preclinical studies.
A significant amount of process re-engineering may be needed to produce a few gramsit is frequently at this stage that they first approach a CMO. For these reasons, mechanical homogenizers, which can be made to operate in a continuous flow mode at any scale, are used at Angel for performing cell lysis of E. coli to release intracellular recombinant protein for purification rather than commonly used laboratory-scale procedures such as sonication or lysozyme treatment.
When proteins are overexpressed in E. coli they frequently accumulate as insoluble inclusion bodies which, after recovery by centrifugation, need to be solubilized and re-folded in a controlled environment to obtain the correctly folded, active form of the protein. At larger scale, controlled-rate dilution into an appropriate renaturation buffer is an option, but dialysis is not. Similarly, buffer changes in lab protocols have traditionally used dialysis or gel permeation, whereas these are either difficult and expensive or sometimes completely impractical to scale up.
Shipston noted that tangential flow diafiltration-ultrafiltration (UF/DF), using hollow fiber or flat sheet membrane cassettes, is the method of choice for large-scale buffer change these days, and it is essential for these strategies to be validated at small scale.
“These are all examples of issues typically encountered during the development of recombinant protein manufacturing processes,” said Shipston, who added that “aggressive timescales rarely take into account the need to perform comprehensive column binding or filtration capacity studies, which are essential to produce a scaleable process.”
Looking at future trends in scale up, Shipston commented that both clients and the CMO can get enormous benefit from implementing single-use technology (disposables) at an early stage in the manufacturing process. The need to validate and routinely perform several cleaning procedures can be eliminated, which speeds up both delivery and turnaround.
Many companies are faced with scale-up issues when they design and build new facilities. Peter van der Meijden, Ph.D., section manager process development microbiology at Diosynth Biotechnology, (www.diosynthbiotechnology.com), described the design, construction, and start-up of Diosynth’s new large-scale upstream production facility, which is dedicated to microbial and cell culture contract manufacturing.
Diosynth, part of Akzo Nobel, has carried out process development including scale up on over 55 recombinant proteins and clinical cGMP manufacturing on over 35. The new plant, comprising 18,000 liters of mammalian cell culture capacity and 10,000 liters of bacterial fermentation capacity, is located at Oss, The Netherlands, and has been designed as a multipurpose, multiproduct facility. Currently, Puregon, a hormone used in fertility treatment, and recombinant human insulin are manufactured on this site.
When it comes to scale up to large volumes, the ability of a cell line to cope with shear forces arising from mixing can be an issue. Diosynth deals with this using scaled-down 60-L bioreactors.
Computational fluid dynamic modeling is then applied to determine where critical spots for high shear force and low mixing will occur. Mixing characteristics can be predicted in combination with a consideration of aeration strategy. “In this way, we test the cell line for its ability to survive under these conditions,” said Dr. van Meijden.
Automation is another important aspect of scale up; it should be used as much as possible in areas such as SIP, CIP, medium preparation, and status checking. However, too much automation results in time-consuming and useless false alarms. It is vital to use only the most foolproof of transmitters and probes to avoid this.
How to run large-scale facilities is also an issue. Currently, the new Diosynth facility is passing through qualification procedures, which has consumed 5,000 man days for the microbial equipment, 3,000 man days for the cell culture equipment, and another 2,000 for qualification of utilities.
Finally, all aspects of scale up are addressed when a company actually turns out commercial quantities of a product. Peter Moestra, Ph.D., divisional vp of biologics manufacturing global pharmaceutical operations at Abbott Laboratories (www.abbott.com), presented some of the financial and technical issues involved in the scale up of Humira.
“Nobody in this industry plans for successbut when the first patient was injected with Humira, the pain went away and there was tremendous pressure on us to produce more,” he said. Humira is an anti-TNF antibody licensed for use in rheumatoid arthritis and psoriatic arthritis. It is also under development for ankylosing spondylitis, psoriasis, and Crohn’s disease, according to the company.
In 2004, sales of the product reached $1.4 billion. The history of Humira goes back to the initial development of a parental cell bank in 1994 and initial production in a 1,000-liter converted chemical process tank. Since this time, the production of the antibody has gone through a rapid sequence of scale ups and process improvements, culminating in a large-scale (12,000-liter) state-of-the art facility in Worcester, MA, where Humira is made by a robust, high-yield fed batch process.
Dr. Moestra explained that going from 3,000 to 6,000 liters involved some minor changes to the growth medium, but going from 6,000 to 12,000 did not involve much change. For scale up of fermentation the company used computer simulation of mixing to detect any potential problems with shear although, it turns out, Humira cells are shear-insensitive. Cell culture curves showed consistency between 3,000, 6,000, and 12,000 production scales.
“We were able to sail smoothly through the comparison studies,” said Dr. Moestra. On lessons learned, he commented, “Success creates demands and you may not find the funding you need in time. For Humira, aggressive investment decisions and process improvements were needed to satisfy demand. Advance planning and experiments can enable process improvement and scale-up to go smoothly.”