Send to printer »

Feature Articles : Jul 1, 2007 (Vol. 27, No. 13)

Designing Better Purification Schemes

Overcoming the Impracticalities of Downstream Single-use Systems
  • Angelo DePalma, Ph.D.

Efforts to raise protein-expression levels have succeeded mightily, but at a cost. Today, high protein titers and accompanying impurity levels have come to dominate downstream processing.

“We have undergone a shift in where bottlenecks occur, from cell culture to purification,” says Jonathan Romero, Ph.D., a senior development engineer at Biogen Idec(

High titer processes are a challenge, especially for clarification and chromatography, notes Jonathan Royce, sterile and clarification group leader at Millipore ( “Some companies solve this problem through innovative technologies that in many cases are proprietary. Others look to more mature industries for solutions, for example flocculation, which has been used for years in water purification.”

Processors handle high-titer process fluids through various techniques, some old and some new. Centrifugation is making a comeback during harvest/clarification, at the expense of ultrafiltration and tangential flow filtration (TFF). Centrifugation is highly suited to high titers and cell densities, and results in lower losses.

During capture, high titers force processors to run multiple cycles, larger columns, or to slow down elution to maximize capture efficiency. Further downstream, concentrated protein solutions, their accompanying impurities, and protein aggregates tax ion exchange and polishing columns. Higher protein concentrations drive down filter flux and cause fouling, necessitating larger filtration surface areas, which in turn affect virus breakthrough behavior. Processors must therefore redesign filtration schemes to minimize membrane fouling. For example, virus filters that run in normal flow-through mode may be re-designed for TFF mode, and the membrane surface chemistries may be adapted to handle the higher flux while still retaining viruses.

The Band-Aid Approach

Companies have thus far used the band-aid approach to overcome these difficulties, solving one problem only to have others crop up, says Dr. Romero. Capture/affinity resins with higher capacity are typical. A complete re-evaluation of the capture step would be better—for example, moving to expanded bed absorption. “At some point, bioprocessors must deal at a high level with the inevitable rise of protein titers to 10 g/L.”

Anything that favorably alters the protein/impurity ratio is helpful. Cells may be engineered to accomplish this, or processors can harvest earlier, with the goal of sacrificing yield of raw protein for final product. “At the end of the day, overall recoverable yield might be higher,” Romero says.

Other suggestions include investigating flocculation for cell harvesting, a technique that might reduce the burden on chromatography columns. Another strategy is to design secondary chromatography steps to operate in flow-through mode—binding impurities, not product—to take advantage of the relative fractions of these two components. Dr. Romero also likes membrane absorption, especially in flow-through mode early in the process, to remove impurities. Finally, eliminating buffer tanks by reconstituting chromatography buffer at the point of use, and eliminating steps between columns (elute, collect, load, repeat) would “save a lot of time and eliminate validation and tankage needed between columns.”

Other experts propose similar measures. Bruno Marques, Ph.D., senior research chemical engineer at Merck & Co. (, believes that reducing reliance on chromatographic separations with nonchromatographic techniques would be valuable as process streams become larger and more concentrated.

“Chromatography offers high resolution but low throughput, especially at larger scales,” he explains. “In the near future, companies will routinely manufacture 100 kg of a mAb per batch. Chromatography is not well-suited for such scales, especially when you consider the buffer requirements of two- or three-chromatography processes.”

Dr. Marques would like to see the development of alternative, simpler, technologies, such as impurity precipitation immediately after harvest. For replacing resin chromatography in flow-through, as well as bind-and-elute, mode, Dr. Marques likes disposable technologies but notes that selectivity and capacity “need further development, especially as far as adsorptive properties and fluid distribution, in the next several years.”

Mimetic replacements for expensive chromatography media are also of great interest. “Whenever we hear about mimetic resins Merck evaluates them, but so far we haven’t found any that meet all our requirements.” Drawbacks of newer mimetic resins include capacity, selectivity, packing, and linear-velocity capabilities.

Single-use equipment now serves nearly every upstream unit operation in one way or another, with single-use bioreactors in the 500–1,000-L range quite common. Even larger processes can benefit from small-volume single-use seed bioreactors and multiple disposable buffer and storage bags of up to approximately 1,000-L working volume each.

Fully disposable, large-scale downstream processing is probably years away, primarily due to the impracticality of single-use chromatography media, equipment (e.g., centrifuges), and certain filtration systems. While single-use direct flow, sterilizing, and certain types of depth filtration are available in disposable format up to the largest scale, disposable tangential flow filtration has only recently been introduced. Pall’s ( Kleenpak™ TFF MF capsule, which debuted at “Interphex 2007” along with disposable fluid management and containers, combines microfiltration with disposability.

According to Jerold Martin, Pall’s senior vp for scientific affairs, Kleenpak TFF resembles a dead-end filter capsule, but is used for cell removal or harvesting. “Kleenpak TFF is much simpler and easier than a cassette system,” Martin says.

Between clarification and polishing, filtration is mostly fully disposable. Pall has developed one fully assembled system whereby harvest fluid passes through depth filter, sterilizing filter, and virus filter, all presterilized and linked through sterile tubing and connectors. This filtration train, which handles process fluids of up to about 500 liters, was custom-designed using off-the-shelf single-use components.

Pall has been working with several larger customers on cassette-based disposable ultrafiltration systems. In these systems all buffer mixing takes place in disposable tank liners and is fed into cassettes, with all the fluid contact components being disposable and supported within stainless steel holders. “This is moving away from where the whole system is disposable, to where there is a fixed platform containing a disposable fluid path,” says Martin. “What people really want is a clean fluid path.”

Polishing and virus filtration can be carried out at almost any scale using presterilized filters connected to collection bags. Pall has created several integrated systems, for example a Mustang Q membrane chromatography module upstream of Ultipor® VF virus filters.

While disposables are rapidly maturing, gaps remain compared to traditional stainless steel manufacturing. “End users and vendors need to work together to identify these gaps and find solutions,” says Royce. One approach is to eschew completely disposable systems for disposable components that work harmoniously with stainless steel. Several vendors, including Pall and Millipore, offer such products.

Chromatography remains an expanding frontier for disposables. The high cost of resins precludes disposable chromatography for high-value products in the near term.

Chromatography: No-Man's Land

Fortunately, membrane chromatography is improving rapidly in terms of chemistry and capacity. Membranes now rival traditional chromatography in performance and capacity for large biomolecules (like IgMs), which cannot penetrate into the pores of traditional media. “Larger molecules lose resolution on traditional ion exchange media,” says Martin.

The same qualities that make membrane chromatography ideal for removal of DNA, host cell proteins, and viruses makes it ideal for large-molecule purification. Pall’s Mustang® Q XT 5000, for example, contains five liters of membrane material, sufficient to purify up to 400 grams of proteins with molecular weights above 400 kD.

Membrane absorbers are gaining in early downstream operations as well, for example in clarification. Membranes will not generate a sludge like a centrifuge, but depending on the process can provide a consistent, high-quality product stream, says Mike Grigus, process engineering manager at GEA Filtration ( Several of GEA’s large industrial biotech projects have used tubular membranes for this purpose.

The move to membrane chromatography is partly based on the desire for less operator involvement in downstream operations. Processors, Grigus reports, are switching from diatomaceous earth (labor-intensive) and centrifuges (high maintenance) to membranes. In addition to convenience and performance, membranes also reduce the opportunity for catastrophic failure.

Ultrafiltration and nanofiltration membranes have experienced evolutionary improvement, for example in quality and molecular weight cutoffs. “There are now more options in the 10- to 2000-kDa cutoff range,” says Grigus, “and more products that work under difficult conditions.” There is also a greater desire to apply membrane technology to more of the post-clarification downstream work.

Downstream improvements are not limited to specific separation products like filters and chromatography media. Improvements that facilitate or streamline process steps can play significant roles. For example, Millipore’s implementation of RFID technology has taken downstream component tracking to a new level.

RFID, which is taking off for track-and-trace in consumer goods and pharmaceutical shipments, helps processors retrieve every bit of information related to a filter, including use, source, properties, and history of integrity testing. Millipore’s SMART RFID technology, for which it has partnered with Belgian firm Tack Smart Filter, was introduced at “Interphex 2007”.

According to Royce, the number of downstream unit operations has shrunk over the last decade. “These process improvements have, of course, led to large increases in productivity, but high protein titers are pushing the limits of the purification steps.”

As part of the drive toward simplicity, bioprocessors are eager to reduce the number of chromatography steps during purification and polishing to two: a protein A capture and a single, disposable anion exchange step. They may achieve this through combinations of protein A, ion exchange, or other resin types but the cost savings derives from reducing chromatography steps, explains Klaus Tarrach, senior product manager for purification technologies at Sartorius Biotech(

Viral clearance is one aspect of manufacturing that cannot be left to Phase III. Consequently, biomanufacturers are implementing strategies for virus inactivation and removal during Phase I and II studies for both enveloped and small non-enveloped viruses.

“With the new European draft guidelines for virus safety in place for investigational biologics, orthogonal strategies for virus clearance should be implemented from an early stage,” says Tarrach.

Ideally, virus filters should provide virus removal independent of their position within the purification scheme and should also be disposable to meet flexibility needs during early process development. Sartorius’ Virosart®CPV accomplishes both, according to the company. Sartorius also offers the disposable UVivatec UVC ultraviolet-C virus inactivation device. This technology is being used for both the inactivation of bioreactor media as well as process intermediates containing mAb’s or recombinant proteins.

Sartorius Biotech is developing smart disposable cellulose-based depth filter products for removing CHO cell-related host cell protein and DNA. When these upcoming products are combined with membrane chromatography products, for example the Sartobind® Q CPV product, processors will have an orthogonal technology platform system for removing cellular contaminants. The two technologies “close the gap between process derived contaminant introduction and DSP-related contaminant removal,” according to Tarrach.