Downstream Bioprocessing: More is More

Feb. 16, 2005
What's good for upstream processes -- higher concentration levels for MAbs and proteins -- can strain downstream operations, forcing manufacturers to rethink the purification train.
A process development specialistat Protherics, Inc. works with aGE Chromaflow column.By Angelo De Palma, Ph.D., Contributing EditorThe dramatic rise in volumetric productivity for cell culture-based manufacturing has been a mixed blessing for bioprocessors. Although desirable from a cost of goods perspective, protein concentrations in the gram-plus per liter range stress the capabilities of processes designed for much lower titers.“High expression levels for MAbs and fusion proteins have taxed purification processes,” says Dr. Brian D. Kelley, director of purification process development at Wyeth Biopharma (Andover, Mass.).
Sidebars for this article:

Downstream Synergies

Transgenics: No Downstream Advantage

Fermentation and cell culture operations occupy approximately the same floor space in a traditional biomanufacturing facility. Tenfold increases in protein titers, which biotech has enjoyed over the past 10 to 15 years, strains the downstream material flow capacity. “What seems like a good thing for the upstream process actually creates new challenges for the downstream process because we’re now handling much more material within the same square footage,” observes Kelley.One consequence has been the need to stage purification of the process stream through multiple cycles for many unit operations, which requires that product be sufficiently stable in the in-process pools being held for the next cycle of operation. Another effect is the need to work at much higher volumes. “In the past it wasn’t unusual to see a hundredfold concentration after capture,” Kelley notes. With protein titers routinely reaching the two to four gram per liter range, concentration factors of ten are more common.Chromatography resins with high protein binding capacities have helped minimize column sizes and in-process pool volumes. Newer resins that bind or capture 50 to 80 milligrams per liter of sorbent have replaced media with binding capacities in the 10 to 30 mg/mL range. Newer resins also tend to be manufactured from less-compressible materials, which withstand high pressures from higher flow rates. These resins (see "Downstream Synergies" sidebar) may be packed to higher bed heights and cleaned more aggressively than some polymer-based sorbents.“Older, more compressible resins didn’t serve large volume processes very well,” says Kelley. “Resin vendors realized that and stepped up to give us many more options.” He adds that current process chromatography equipment and resins should keep up with the increased titers for a few years more, until the titers exceed ten grams per liter.Initial cell removal and clarification steps are also burdened by the elevated cell densities and extended culture times commonly employed to raise protein titers. Processors have therefore migrated from filtration toward continuous centrifugation for initial clarification of mammalian cell cultures. Centrifugation was originally thought to be too harsh for mammalian cells, but this is not the case for newer centrifuges with carefully designed inlet distributors.Centrifugation produces feedstreams that require more polishing compared to streams subjected to microfiltration, which increases the need for high-capacity depth filters downstream of the clarification step. “There’s been a steady evolution and competition among major filtration vendors over the last five years, but cleaning up harvest feedstreams still represents a challenge because of our high debris load which needs to be removed to protect chromatography columns,” says Kelley.Cell culture’s high productivity has caused a shift in the cost of goods equation from fermentation to purification. Wyeth Biopharma has responded with initiatives on process intensification, yield maximization, and minimizing amounts and numbers of buffers. “Our simplest processes are the most elegant,” says Dr. Jeffrey Deetz, Wyeth’s senior director of drug substance development, “but they take the most time to develop while maintaining the usual goals of purity, product profile and quality.”On the development side, Wyeth has adopted a platform approach, currently a two-column process. Engineers use a 96-well, plate-based, high-throughput screening method to optimize washes, salt concentrations, pH, buffer excipients and other performance-enhancing parameters for the two columns. “If we can do a quick survey and not hold up the critical evaluation of biopharmaceutical candidates, then we can tailor at least some aspects of the purification train to give a better product profile for early evaluations,” says Deetz. “At the same time we don’t want to over-invest in all these candidates because so many are coming through the pipeline.”Orthogonal operationsDemonstrating viral clearance will always be part of mammalian cell-based downstream processing. The trend from animal-derived media components to completely defined media will reduce the viral clearance validation burden somewhat, since processors need only worry about infectious agents from cultured cells and not from media components. However, the virus-generating potential from protein-producing cells will exist as long as such cells are used.Dedicated viral clearance filtration is usually deployed as far downstream as possible due to the membranes’ tendency to foul and clog. “Filter fouling can seriously affect process economics,” says Dr. Jeff Carter, consulting product manager at Millipore Corp. (Billerica, Mass.). In practice, unit operations provide multiple, orthogonal operations to achieve clearances on the order of 12 to 18 log. Virus-clearing capability is built into filtration and chromatography. Additional filtration or inactivation steps are added as needed to reduce virus levels to FDA/EMEA standards.Most dedicated viral clearance methods involve tradeoffs. For example dead-end filtration often results in unacceptable protein losses. “Filters clog, involve long processing times, and sometimes break through,” notes Maik Jornitz, group vice president VP for bioprocess at Sartorius North America (Edgewood, N.Y.).Chromatography is useful where similarly-sized viruses and target proteins might get caught up together in a dead-end filter. However, chromatography is extremely expensive and its primary purpose is purification, not virus removal. Validating virus retention and/or elution through columns adds an unnecessary step to the viral clearance exercise. Chemical inactivation works well enough, especially when combined with orthogonal methods, but removing the spent chemical treatment adds a processing step.Sartorius has developed a three-pronged orthogonal approach to viral clearance that appears to minimize these negatives. The technique employs three steps: membrane adsorption, ultraviolet-C inactivation, and filtration with a 20-nanometer dead-end filter.Key to the process are Sartobind Q anionic exchange membrane adsorbers. Think of these as a cross between filter membranes and chromatography media, with the speed of the former and the selectivity of the latter. In one study Sartorius investigators, working with Abgenix (Fremont, Calif.), achieved log removals of a typical panel of test viruses ranging from 4.41 (for a very high-dose minute virus of mice) to 7.53 (reovirus-3). These results are comparable to those for running a separate anion exchange column. The membrane adsorber, however, consumes considerably less buffer, takes less time and is less expensive.Capturing value in MAb purificationThe first step in monoclonal antibody (MAb) purification is “capture” on a protein A chromatography column. Derived from a bacterial toxin, protein A can generate nearly homogeneous MAb in a single step. Protein A sorbents are expensive (as much as $12,000 per liter) and difficult to clean. Still, they remain the sorbent of choice for MAb manufacture.Several vendors have introduced “artificial” protein A (or protein A-mimetic) sorbents. ProMetic (Montreal) claims its MAbsorbent chromatography sorbent binds MAbs as efficiently as protein A media.Protherics (Brentwood, Tenn.), which develops emergency and oncology biotherapeutics, recently re-designed its purification method for its Crofab snake venom antibody to exploit protein A-mimetic alternatives. Originally the process employed a sodium sulfate precipitation of sheep immunoglobulins, followed by a “very long” purification process, according to process development director Richard Francis. The new method uses Prometic’s protein A-mimetic resin. Sheep serum is applied directly to the column and eluted at pH 3. Resulting protein elutes at 98% purity and 95% isolated yield. In addition to MAb binding comparable to that of protein A, the Prometic resin provides viral clearance.Pall’s alternative sorbents, MEP HyperCel and MBI HyperCel carry immunoglobulin-selective ligands, cost about one-third as much as protein A media, and are cleanable with caustic (validated to 200 cycles). Dr. Warren Schwartz, senior technical director for Pall’s chromatography products unit, claims immunoglobulin affinities for the two products, 20 to 40 mg/L, are comparable to that of protein A.MEP HyperCel sorbent is currently employed in purification of ten drug candidates in clinical trials, including both antibody and non-antibody proteins.
Downstream Synergies

As biotech matures, biochemistry is giving way to chemical engineering as a strategy for reducing cost of goods, says Dr. Fred Mann, technical marketing manager for chromatography at Millipore Corp. (Billerica, Mass.). “We’re seeing a lot more interest in process scheduling, process compression and integration or combination of process steps,” he says.

Higher protein titers place increasing demands on chromatographers, not necessarily for higher capacity resins but for overall improved productivity. “Processors looking to purify an entire batch in one shift must consider a resin’s flow characteristics,” says Mann. Lower-flow columns take more time, which may compromise protein stability.

Traditionally, chromatography resin was selected at early stage laboratory development and taken all the way through the process as it scaled up. Today, Mann sees greater readiness to change, as processes are optimized for full-scale manufacturing and reduced operating costs.

Pall Corp.’s (East Hills, N.Y.) November 2004 acquisition of Ciphergen’s BioSepra division dovetails with Pall’s strategic alliance with Euroflow, which manufactures chromatography columns for pilot- and production-scale applications of Pall’s PK series of chromatography process systems.

Scalability and support for robust processing have long been a goal of bioprocess chromatography. Historically, sorbents were based on macroporous polymeric beads, particularly crosslinked agarose (GE Healthcare) or polymethacrylate (Tosoh Biosep). However, bead swelling and shrinkage with changing fluid flow, pH and ionic strength disrupt bed integrity and impair performance.

Several vendors now offer sorbents that combine absolute rigidity with the desirable chromatographic properties of soft hydrogels. One line of Pall’s sorbents is formulated from rigid, large-pore mineral composites combined with hydrogels. Monomers are infused into the pores of the mineral composite and polymerized in situ. The hydrogels provide high binding capacity and fast mass-transfer while the mineral-composite confers rigidity.

With diameters typically ranging from 20 to 200 cm, packing a production-scale column consumes three to eight hours and requires three or more operators. Success depends on the skill and experience of the operators, particularly as scale increases. If the packed column does not meet performance specifications, the entire process is repeated.

Euroflow columns are slurry-packed using an automated packing system that provides reproducibly-packed columns, relieving the need to handle and maneuver heavy column-inlet assemblies. Packing and unpacking operations are inherently safer and can be accomplished by fewer operators. Pall Ceramic HyperD and HyperCel sorbents have been shown to operate well in Euroflow columns. So, the quest for enhanced process economics is supported by both sorbents and columns.

Transgenics: No Downstream Advantage

Transgenic plants as pharmaceutical production organisms promise to streamline upstream operations through lower capital investments, easy scaleup, and low production costs for raw material. “Transgenics can beat mammalian cell culture every time,” says Prof. Zivko Nikolov at Texas A&M University (College Station, Texas). Raw protein from CHO cells costs between $100 and $300 per gram, whereas plant-derived proteins are in the $10 to $15 range.

Unfortunately, transgenics firms are discovering that those benefits don’t translate into similar benefits downstream.

“Developers thought originally that downstream benefits would be similar, but they’re not,” Nikolov laments. “Sure, we don’t have to do viral clearance studies and related validation, but that’s probably the only substantial difference.”

Seed and leafy crops raise distinct but equally complex downstream complications. Plant leaves contain fewer protein contaminants than do cell lysates, but sequester nasty alkaloids, pigments, and phenolics, some of which are toxic and all of which must be removed.

That being said, the cost of goods allocation for upstream processing is still a hefty 40% to 50% of the total. Saving half of more of that number provides substantial improvement in cost of goods.

However, the future of transgenic pharmaceutical-producing plants is still murky. It’s apparent now that field crops are out due to fears of “killer corn.” Even self-pollinating plants like rice will probably not pass regulatory muster, says Nikolov. “Plus, there will be more stringent regulation on growing drugs in food crops. All of this will contribute to upstream and overall costs.”

Plant-derived proteins are not succeeding because biotech companies aren’t buying into the idea, says Nikolov. “The pioneers came from ag-biotech and brought a seed company mentality along,” he says. “To them, expression levels are everything, but they don’t have an integrated process yet that addresses upstream and downstream concerns. Interesting a top-tier biotech company would take a dramatic improvement in cost of goods, and they’re not there yet.”