Send to printer »

Feature Articles : May 1, 2011 (Vol. 31, No. 9)

Scale-Up Woes Hinder Protein Purification

Costly Methods Drive Search for Streamlined and Efficient Alternatives to Current Practice
  • Angelo DePalma, Ph.D.

Depending on the process and who provides the statistics, downstream processing accounts for between 50–80% of the cost of producing biotech drugs. Whatever the exact number, protein purification looms large in the economic viability of a product. Hence the interest in systematizing and streamlining recovery and purification, from discovery scale to multiton production.

Marcel Ottens, Ph.D., assistant professor in the department of biotechnology at Delft University of Technology, sees growing interest in QbD and design of experiment (DOE) approaches that incorporate process analytic technology.

Dr. Ottens, who specializes in high-throughput experimental and mathematical approaches to optimizing chromatographic separations, predicts ongoing improvement in higher-flow and -capacity chromatographic resins, membranes, and monoliths, innovation in ligand chemistry (e.g., mixed-mode resins), affinity antibodies, antibody fragment-based chemistries, and greater reliance on operational modes like continuous and simulated moving-bed chromatography. Augmenting these advances will be greater adoption of disposables and an emphasis on facility fit to accommodate higher expression levels.

Shifts in expression systems with direct impact on purification will occur as well. “Although mAbs produced via mammalian cell culture currently dominate the protein therapeutic market, simpler expression systems like yeast and E. coli will become more significant,” he says.

Despite these innovations, Dr. Ottens predicts that the industry’s innate conservatism, forged by regulatory issues, will continue to hamper process improvements. Other hurdles include the high cost of capture resins, higher titers, and the complexity of therapeutic biomolecules. “We still lack sufficient knowledge of the physical-chemical properties and phase behavior of proteins, particularly with respect to adequate purification process design.”

Integrated Approach

Recognizing these changes are coming, vendors are viewing separation scale as a continuum. As QbD concepts infiltrate protein development and manufacturing, small-scale and early-stage protein isolation will be viewed as a warm-up for pilot and large-scale purification. DOE is already helping processors identify binding and elution characteristics with an eye toward scaleup.

GE Healthcare’s recent introduction of PreDictor™ plates and the ÄKTA™ avant line of chromatography systems exemplify this.

PreDictor Plates, which are 96-well microtiter plates charged with resins for rapid screening of conditions critical to binding, elution, and washing, show excellent correlation with chromatography column results.

From these miniaturized experiments process developers can move on to PreDictor RoboColumn™, consisting of miniature columns packed with resin. Another option is the column-based ÄKTA avant instruments, which help fine-tune conditions derived from PreDictor plates in small-column formats and provide robustness testing.

“Together, these products help process scientists characterize their design space efficiently, and thereby, develop a safe and robust process,” comments Eric Grund, Ph.D., senior director, biopharma applications. “DOE software used with these products allows you to do fewer experiments and get better understanding of the process before scaling up.”

Dr. Grund recognizes that it’s possible to overthink and overcharacterize processes before scaling up. “You have to be smart to know when to move on to the next study. We’re still all learning how to use these tools.”

Novel antibody formats like fragments and IgMs that lack Fc regions are changing the notion of platform separations. At the same time, therapeutic molecules are becoming more complex, for example biobetters with extended plasma half life, cell culture-manufactured vaccines, virus-like particles, and other molecules for which platform separations do not exist.

“If you don’t need an Fc region, why produce a 120 KDa molecule when you can achieve the same result with an antibody fragment?” asks Laurens Sierkstra, CEO of BAC, a specialist in high-affinity protein purification products.

BAC in essence creates purification platforms for complex molecules by designing protein A-like affinity resins for initial capture from fermentation broth, cell culture, or plasma. Other steps, such as ion exchange, polishing, ultrafiltration, and virus inactivation are performed as needed.

With GE Healthcare, BAC has introduced products for purification during discovery and preclinical stages through large-scale production: KappaSelect for antibody fragments, VIIISelect for therapeutic proteins, and AAV for viruses. “We are building a product portfolio that can act as the ‘protein A’ for the purification of all major classes of new biotherapeutics,” Sierkstra says.

The technology is based on single-domain antibodies, which are 110-residue peptides consisting of one variable domain of a heavy-chain antibody. Single-domain antibodies exhibit the same affinity as intact antibodies but are more resistant to heat and cleaning regimens. The antibodies are generated in llamas then cloned and expressed in large quantity in baker’s yeast. Protein A is produced similarly in E. coli. The ligands are attached to a suitable bead or surface to form the capture resin. According to Sierkstra, resulting resins are stable to base and strong acid.

Facility as a Purification Asset?

Inese Lowenstein, director for downstream processing at EMD Millipore, notes that as titers continue to rise, mAb producers face difficulties in maximizing facility utilization and avoiding high-capital investments. “Titers constrain the filtration area, buffer tanks, and hold-tank volumes—the entire downstream facility.”

Millipore has, therefore, focused on developing high-capacity products such as ProSep Ultra Plus, a protein A affinity medium that, like GE’s recently introduced MabSelectSure LX, achieves high throughput while reducing column sizes, buffer volumes, and number of purification cycles.

Lowenstein cautions against viewing high-productivity unit operations in isolation. Their greatest benefit, she says, is in their integration into a process that reduces buffer-related steps and the need for some or all holding tanks. Millipore has presented a three-step protein-purification process that achieves these goals: protein elutes directly from the protein A column to an Eshmuno S (cation-exchange) column, and finally through an anion-exchange membrane adsorber for polishing.

“The process reduces time and effort, minimizes precipitation, has a smaller footprint than conventional purifications, and better utilizes facility resources.”

Contract manufacturers are by nature even more risk-averse than innovator companies. The emphasis at CMOs is getting the job done as quickly and cost-effectively as possible. Therefore CMOs rarely stray from their clients’ idea of robust purification methodology.

Scott Kelly, manager for GMP manufacturing sciences at Paragon Bioservices, notes that disposables have made a significant difference in providing value for early downstream operations like recovery and clarification.

“The productivity of upstream operations means more protein, but also more cell debris and side products.” Removing those impurities before chromatography is a key element of success.

Utilization of space, equipment, and personnel become bottlenecks as processes grow in size. Processes that employ centrifugation at bench scale must eventually switch to tangential flow or depth filtration, which adds time and equipment. “You need staff to prepare those buffers, and footprint to prepare and store them. So the more efficient we can make recovery and clarification the better,” Kelly says.

Similarly, large-scale batches need to be split and held, with several capture cycles required per batch instead of one. Here, not only floor space, but resin capacities come into play. “You have to fit the project around what you’re capable of at any particular time.”

Competition

Since designing ease of purification into molecules is difficult, bioprocessors will increasingly turn to nonmammalian expression systems with a downstream advantage such as fewer steps or improved product quality.

For example Ajinomoto’s Corynebacterium-based expression system, Corynex, reportedly reduces costs and speeds time to market by expressing recombinant proteins directly into broth. Corynex works with cytoplasmic proteins, growth factors, enzymes, and mAb fragments, but not with membrane proteins or entire antibodies.

The elimination of lysis minimizes host-cell proteins, DNA, and other contaminants, with an additional bonus. “Corynex allows

purification of properly folded proteins directly from the broth, with no inclusion-body processing,” says Joel White, business manager for biotechnology. “In the case of IGF-1, a protein with three disulfide bonds, purification requires half as many steps as conventional bacterial fermentation.” By eliminating junk and folding proteins correctly, losses are reduced to half of what one would expect with E. coli.

Purification involves centrifuging the cells away, ion-exchange or reverse-phase chromatography, followed by gel filtration or ultrafiltration. Eliminated are harvest, disruption, removal of cell debris, and refolding.

Insect cells are viewed as a simpler alternative to mammalian cell culture, perhaps providing a purification advantage as well, but problems still need to be ironed out.

In many respects insect cell culture is at the stage at which mammalian cultures were about a dozen years ago. Cell densities barely reach 2.5 million cells/mL, limited by the necessity that cells be in their logarithmic growth phase during transfection. Titers are below average for many established protein processes, and quite low for mAb processes. According to Manon Cox, Ph.D., president of Protein Sciences, one can expect “a couple of hundred milligrams per liter” from insect systems.

“But a lot of productivity is still to be gained as soon as the first product produced in insect cells is approved,” says Dr. Cox. “At that point, media development and feed strategies will take off,” perhaps providing some of the dramatic improvements we’ve seen for CHO cells.

Advantages of insect culture include proper protein folding and rapid, transient expression. Baculovirus-based products are more immunogenic than those from more conventional expression systems, which may be advantageous for vaccines but not for other products. Another minus is the presence of substantial quantities of baculovirus, which—even though they are not pathogenic to humans—must be cleared.

But perhaps the greatest drawback is regulatory uncertainty. “We filed our license application for an influenza vaccine three years ago,” Dr. Cox says, “and it is still not approved. All the problems are substrate related.”

Where It All Starts

Downstream bottlenecks are not the exclusive domain of large-scale processors. Initial or proof-of-principle purifications, the training ground of sorts for production scale, must often start from scratch, particularly for contract R&D firms.

Outsourcing early-stage purification and expression is becoming popular due to the increasing number of competitive service providers, says Brett Modrell, director of process development at Monserate Biotechnology. Monserate, which specializes in batches of up to about 10 g, operates in the purification “trenches”—on molecules whose total purification portfolio may include little more than an HPLC trace.

Monserate first considers formulation and purification possibilities based on solubility and activity. At this stage, the major bottleneck is not too much protein straining facilities and equipment but the more prosaic lack of it. “The greater the expression levels, the easier the purification,” Modrell says. “Effort spent improving expression pays back many times over.”

Yet even here operations such as sample concentration and buffer exchange become as critical as they are for larger processes. One notable difference at this scale is the potential for using affinity tags based on transition metal chelates, which provides enough protein to characterize, and perhaps to experiment with. Affinity tags are impractical and undesirable at production scale.

Similarly, chromatography media chemistries, ligand density, capacity, and cleanability are characteristics as desirable at 5 g as 50 kg scales, as are the benefits of disposable mixing and fluid handling.

Eventually, if all goes well and the molecule progresses through development, the fruits of lessons learned at milligram and low-gram scales find their way to process development.