But, alas, regulatory uncertainty can make cowards of us all. Demonstrating product equivalence late in development is so costly and risky that many successful biotech companies "just say no"--at least publicly--to downstream process innovation.
Future downstream process innovations will probably trickle up into the conservative mainstream from nontraditional biotech areas such as transgenics and gene therapy, as well as from the plasma products industry, which predates fermentation-based biotechnology. But for now, downstream innovations primarily build on decades-old unit operations. Nevertheless, many of these technology tweaks are reducing manufacturing costs and speeding up commercialization, allowing manufacturers to achieve larger economies of scale or to clear regulatory hurdles faster.
Throw It All Away
Because of cleaning costs and associated validation issues, disposable purification equipment often appeals to engineer and accountant alike. From filters and housings to chromatography columns and fittings, disposables are sterile out of the box, eliminate product cross-contamination, and best of all, require no cleaning or cleaning validation.
Throw-away equipment is even more appealing when augmented by disposable connectors, valves, and piping. Leading filter and chromatography vendors, including Pall Corp. (East Hills, N.Y.), Sartorius (Edgewood, N.Y.), Millipore Corp. (Billerica, Mass.) and Meissner Filtration (Camarillo, Calif.), are moving headlong into disposables whenever the economics warrant it.
Completely disposable downstream processes have not yet hit the big time for bioprocessors, but pieces of the disposability puzzle are filling in. Combinations of disposable filtration systems, chromatography, and concentration/desalting are available for medium-sized and smaller processes.
"The decision to go disposable is usually based on the value of product, cost of manufacturing, and, to a certain extent, overhead costs of facility and labor,"says Pall's Derek Pendlebury, Ph.D. "Fifteen years ago everyone we talked to said --great idea, not interested. But increased understanding of true biomanufacturing costs has changed their view," Pendlebury says.
Traditionally, purification was be-lieved to make up half of biomanufacturing outlays, assumed to be mainly material and consumables costs. Today, drug-makers also factor in the costs for infrastructure and personnel, which shifts the center of gravity for costs slightly upstream. System time accounted for 54% of total manufacturing costs for one of Pall's European contract manufacturing customers. Of remaining costs, 36% came from materials and 10% from personnel.
To cut its overall process costs, the company focused on reducing system time. It could either lower per-batch costs by increasing the number of batches, or increase batch size. Because sponsors usually fix batch size, increasing throughput was the only remaining option. And because 64% of its manufacturing costs were already fixed, the company decided to increase the number of batches it could process by deploying disposable purification equipment wherever possible.
Chromatography Meets Filtration
Combined unit operations, such as centrifugation and filtration, are improving downstream bioprocess operations. However, hybrid technologies also are taking shape, notably membrane chromatography (MC), which combines filtration and chemical affinity in a single device.
MC, which uses filter media modified with typical chromatography chemistries such as ion-exchange groups, is replacing column chromatography in some niche applications such as final polishing and virus/endotoxin removal. Because membranes are thin, companies stack them to achieve desired capacity. Newer formats such as spiral wound membranes are another way to achieve higher specific capacity.
Lately, MC has begun to take on the look and feel of traditional column chromatography. Sartorius, for example, now specifies its MC products by bed volume, just as chromatography vendors have done for decades. More significantly, bioprocessors have stretched MC's protein-binding capacity beyond polishing to medium-volume bioseparations.
Although MC?s capacity still doesn't touch that of classical media, multiplexed MC media offers surprisingly robust ---and lightning-fast---separations.MC and gel chromatography are complementary, says Maik Jornitz, group vice president at Sartorius North America. "Processors appreciate the combination for its simplicity, speed and reusability," he says. Typically, users run MC cartridges in parallel for scaleup, then serially to obtain a sharper breakthrough curve, according to Jornitz. MC media resist degradation at up to 1,000 cycles and are also easier to clean than chromatography gels.
What they lack in capacity, chromatography membranes make up for in speed, especially for removing low-concentration contaminants. For example, Pall's Posidyne membranes are functionalized with cation exchangers for removing endotoxins.
Mustang, another Pall membrane chromatography product, uses low protein binding polyether sulfone backbones and a higher cationic exchanger loading for protein polishing and purification of oligonucleotides. The membranes offer potential savings in complex bioprocesses such as conjugate vaccine manufacturing.
Old Dogs, New Tricks
Long the workhorse of bioseparation, chromatography also is getting its share of process improvements. Traditional gel media, for example, are notoriously difficult to scale up: as columns become taller, huge pressure drops cause gels to compress, changing the physics of media adsorption.
"The only way to scale them up is by increasing diameter,"says Robin Rogers, Ph.D., professor of chemistry at the University of Alabama. "Eventually it becomes impossible to guarantee a constant pressure-drop distribution in the radial direction, and resolution suffers." At that point, says Rogers, biomanufacturers are forced to add chromatography units, adding complexity to the process.
Rigid media that can't be bent, folded, or mutilated can overcome many of the mechanical limitations of soft gels. For example, Millipore's ProSep-A HC chromatography media use a mechanically rigid controlled-pore glass base material rather than a compressible gel, faithfully reproducing flow and adsorption over a wider range of pressures than do polymer-based gels.
For capacity-hungry products like monoclonal antibodies, cost and competitive pressures may render traditional chromatography more of a luxury than a necessity. Transgenics, particularly engineered plants, are changing the way biotech looks at downstream processing.
By necessity, agricultural biotech firms grow and process protein-producing maize, rice, potatoes, and other crops like foods, due to the huge biomass generated from harvesting tons of plants. Because plant-produced proteins are expressed in specific tissues, though, downstream operations are simplified: protein capture almost always employs simple chemical/food operations such as extraction or centrifugation.
For example, Sembiosys (Calgary, Alberta) expresses therapeutic proteins in the oil bodies of safflower seeds and isolates proteins through continuous centrifugation. "By addressing the entire manufacturing issue, we've changed the economics of protein purification,"says CEO Andrew Baum. SemBioSys calls its process StratoSome.
To make injectible formulations, SemBioSys cleaves proteins from oil bodies, centrifuges, then follows with column chromatography. But when the oilbody assembly itself is the product, say for topical formulations, "you're done," Baum beams. A related oil body technology, StrataCapture purification uses affinity-modified oilbodies to purify proteins generated through more conventional expression systems.
Extraction Underutilized
Extraction, although a mainstay of chemical and food processing, has yet to gain ground in bioprocessing. Although some companies, including Genentech, have published papers on protein extraction, most work in this area today occurs at universities. It's safe to say that at commercial biotech ventures, chromatography operations outnumber extractions by a huge margin.
This is surprising, given the relative costs of chromatography and extraction. Conventional chromatography media cost hundreds of dollars per liter and specialty media even more; in extraction, the "disposable" is the solvent. Extraction offers several additional advantages over chromatographic capture: rapid mass transfer, easy scale-up, very low cost, and tolerance of dirty process fluids. In addition, extractions can separate cell debris from protein, eliminating an initial filtration/clarification step.
So why the resistance? According to University of Alabama chemistry professor Robin Rogers, the initial design of extraction processes is more difficult than it is for chromatography. "It is an engineer's process, not something that can be bought off the shelf like chromatography," he says. "Plus, biochemists understand chromatography."
Bioprocessors undoubtedly will rethink simple unit operations, like extraction, as processes and material requirements for new protein drugs grow. "There is real concern that current technology will not meet the demand for the new generation of protein therapeutics which will be required in gram quantities per treatment," says Rogers.
SemBioSys's StrataCapture is essentially a technique for dissolving normally water-soluble proteins in an oily matrix, followed by extraction.
Aqueous two-phase extraction (ATPE) using water and saturated solutions of polyethylene glycol (PEG) and salt, PEG/carbohydrate, PEG/copolymer, water/salt/alcohol, and others, can separate proteins from other organic materials and cellular debris. For example, Mike Zhang at Virginia Tech uses ATPE, to obtain 87% pure egg white lysozyme from model protein/tobacco leaf homogenates. He achieves this using a simple partitioning between pure water and water spiked with polyethylene glycol (PEG), dextrin, and salts. Zhang never knows which phase will take up protein, but as long as one phase does and the other doesn't, the process promises a simpler, less expensive alternative to traditional protein purification.
Precipitation Shows Promise
DNA products present both technical and philosophical hurdles. Bioprocessors who work with proteins treat DNA as a contaminant; when DNA is the product, protein becomes the contaminant.
Chromatography works well enough with the coiled plasmid DNA used in DNA vaccines. However, plasmid DNA bioavailability is poor, which means doses ---and manufacturing batches ---tend to be large. The high cost and unreliable specificity of chromatography for plasmid vs. other forms of DNA prompted Russ Lander, Ph.D. of Merck (West Point, Pa.) to look for a faster, less expensive way to purify plasmids. Lander discovered that cetyltrimethylammonium bromide (CTAB), a cationic detergent, selectively precipitates plasmids while keeping proteins, RNA, and endotoxins in solution.
Standard plasmid purification in-volves anion exchange chromatography followed by reverse-phase chromatography. "We've replaced those two relatively expensive operations with an inexpensive, classical unit operation." CTAB also achieves several-log endotoxin removal and sizeable clearance of endotoxin and RNA. "[The] detergent is like a surgeon," says Lander. "We've really hit the jackpot with this technique."
Even with extraction replacing chromatography, plasmids present unique downstream hurdles. Plasmids are shear-sensitive, so homogenizers for cell lysis are out of the question. "You have to use chemical methods," says Lander. High-speed centrifuges, rapidly moving feed zones and over-zealous pumping also are out.
After precipitation, plasmids are polished using a selective silica adsorbent, LRA, from Advanced Minerals (Goleta, Calif.). Lander admits that he was as lucky with this technique as he was with CTAB. He'[d expected the silicate simply to soak up endotoxin but, he found that it removed a host of other impurities, polishing away the remnants of host DNA and other plasmid preparation contaminants.
Tips for Ensuring Adequate Pathogen Removal With viral clearance built into most bioseparations, bioprocessors must nevertheless assure adequate clearance of model viruses from all mammalian cell culture production batches. Most purification trains, especially those that are chromatography-intensive, clear enough virus to satisfy regulators, but lapses still occur. Katherine Bergmann, Ph.D., viral clearance study director at Charles River Laboratories (Wilmington, Mass.), offers the following advice:Design purification with viral clearance in mind. Adopt several orthogonal viral clearance/inactivation steps that work through different mechanisms, including at least one chemical inactivation. Plan ahead for material needed for viral clearance and general process validation. Be prepared for the effects of clearing less virus during scaledown: Since there is less virus to clear, lower nominal viral clearance will be obtained. Assure that spiked viruses do not clog nanofilters during validation studies in ways not observed during normal processing. Evaluate cytotoxicity or interference from process components before validation studies because some process components, including preservatives, interfere with viral clearance readings. Remember that Q-PCR only measures clearance, not viral inactivation. Viruses are not the only pathogens of concern. "Mad cow" and related diseases raise issues over prion contamination in all things bovine, from meat to media, which includes serum albumin and dozens of cosmetic and pharmaceutical products. As with viruses, prion clearance validation involves spiking a scaled-down process with infectious agent and demonstrating that prion protein has either been removed or is inactivated. In May, 2003, Bayer Biological Products (Research Triangle Park, N.C.) received patent allowance covering the use of a proprietary, prion-specific Western blot assay to confirm prion removal from human- and bovine-derived bioproducts. The technology, which was licensed to BioReliance (Rockville, Md.), does not provide a blueprint for removing prions---that is up to individual processors. However, it does offer a measure of security against products transmitting such frightful maladies as Gerstmann-Straussler-Scheinker syndrome, Kuru, and fatal familial insomnia.Protein A Mimics for Antibody CaptureMonoclonal antibodies (MAbs) are the fastest-growing segment among biopharmaceuticals, as the success of Humira (Abbott; rheumatoid arthritis), Raptiva (XOMA/Genentech; psoriasis), and Rituxan (Genentech; oncology) will attest. MAbs are relatively high-dose drugs administered chronically or long-term, which means that a lot of material is needed to meet demand. Protein capture, the initial step in most MAb downstream processes, uses Protein A affinity media to remove MAbs from protein, DNA, viruses and other impurities. Protein A naturally binds to mammalian immunoglobulins (antibodies), which is exploited in a variety of commercial protein A capture media. Millipore claims its ProSep-A HC Protein A media purifies more MAbs than any other product. Other vendors, including Sterogene and Amersham Biosciences, offer variations on this theme. Protein A can generate 95% pure protein in one step, even from dirty feedstock, while providing decent levels of viral clearance, but it has several drawbacks. At $10,000 to $12,000 per liter (vs. hundreds of dollars for ion exchange), the media aren't for tightwads. Other drawbacks include ligand leakage, species-dependence, low-pH elution which might damage some MAbs, and difficult cleaning/reuse.Fremont, Calif. based Ciphergen and ProMetic of Montreal are currently developing media that mimic Protein A, promising lower cost, easier cleaning, and greater reusability. Ciphergen's HyperCel hydrophobic charge induction chromatography product line captures MAbs through a low-cost, dual-mode ligand. MEP (mercaptoethyl pyridine) HyperCel attaches to proteins through mild hydrophobic interactions, and selects MAbs through interaction with divalent sulfur close to the ligand's pyridine ring. Ciphergen claims single-pass IgG purities of 88-90% and recoveries as high as 87% with MEP. Next year Ciphergen will introduce MBI (mercapto-benzimidazole-sulfonic acid) affinity ligands, which promises to weed out non-MAb proteins from plasma sources.The idea behind ProMetic's (Montreal) MAbsorbent synthetic affinity ligand adsorbents is similar. Developed using computer modeling to mimic IgG binding sites, MAbsorbent ligands work with all polyclonal and monoclonal IgG sub-classes. At a protein loading rating of 40 mg per mL of settled gel, Prometic claims this ligand is as efficient as Protein A. Prometic marketing manager Victor Bornsztejn would not give specifics on costs of synthetic ligands vs. Protein A. But when media, buffer usage, disposables and recycling were considered along with media costs, he admits that MAbsorbent was "decidedly less expensive---well under half the cost."
