May 15, 2015 (Vol. 35, No. 10)

Angelo DePalma Ph.D. Writer GEN

Bioprocessors Might Grapple With Contaminants, Media Formulations, or Dissolved Oxygen Fluctuations, or Even Edge toward Continuous Processing

Last month’s Bioprocess International European Summit in Düsseldorf featured a number of talks on state-of-the-art techniques for improving cell culture. The presentations were particularly noteworthy in providing detailed advice on how to meet the specific challenges of carrying out successful cell culture operations.

For example, mycoplasma contaminations are one of the most devastating cell culture events. At around 0.2 microns in size, the organisms are the smallest bacteria known. Passing process fluids through 0.1-micron sterilizing grade filters should remove mycoplasma, but the organisms’ changing size and shape lets them slip through filter membranes under certain condition.

Nick Hutchinson, market development manager at Parker Hannifin Manufacturing, advocates risk-based assessment and mitigation to identify and rank potential sources of mycoplasma contaminations. He adds that contamination risks should be weighed against mitigation costs.

This approach is sound and even qualifies as a “best practice” in other biopharmaceutical operations, but it is not strictly quantitative. Individuals rank risk differently, and mycoplasma contaminations are rare.
“You cannot easily obtain data from which to derive probabilities,” Hutchinson notes. Instead bioprocessors rely on historical and literature data.

Since mycoplasma can elude even sterile filtration, that operation is one clearly identifiable source of risk. Hutchinson’s team discovered that virus breakthrough correlates strongly with pressure drop across the membrane: The lower the pressure, the better the retention. Using the modified risk calculation of exposure-times-danger-times-cost, manufacturers must then decide whether the added process time will benefit their process.

“It is hard to generalize what that lower pressure should be,” Hutchinson admits. “It is worth studying where the cutoff point may be—where you start to see virus breakthrough. Whatever you decide the cutoff point is, you cannot apply it to every medium, organism, and filter.”

A development team at Parker Hannifin Manufacturing studied virus breakthrough across filter membranes and determined that mycoplasma retention falls off rapidly at pressure drops above 0.4 barg. Pressure values, however, may differ significantly from one process to another.

Chemically Defined Media

Concentrations of every component in a chemically defined medium are known precisely, enabling manufacturers to duplicate the exact formulation time after time. But variability still exists in the form of impurities and trace components.

If bioprocess scientists have sufficient resources, they can analyze—and potentially control—every component for impurities. “Most formulations don’t require that unless you’re troubleshooting,” observes Elizabeth Dodson, Ph.D., research and development manager, advanced bioprocessing unit, BD Biosciences.

Such depth of scrutiny is impossible with complex formulations—animal- or plant-derived peptones—because of their biological origins. Depending on how one counts, peptones may contain tens of thousands of components including peptides, metabolites, and metals. Bioprocess scientists and manufacturers can nevertheless test for consistency, for example, in molecular weight distribution, amino acid composition, and metals. But natural variation will always exist.

The analytical range for complex media can be as high as 20%, compared with less than 1% for chemically defined products. “You’re going to get that type of bounce due to biological variability,” Dr. Dodson points out. “But while you lose absolute consistency, you gain the synergy of a biologically deeper formulation. These complexes work synergistically in ways that are not currently captured by defined formulations.”

While peptones may provide a good level of similarity batch to batch, products from different suppliers show a great deal more variability, even within the same general class (such as yeast hydrolysate). How does one decide which medium is best? Dr. Dodson advises that one should first understand process drivers, the precise components that affect product yield and quality. “If process developers lack information on those,” explains Dr. Dodson, “they need to screen across several peptones and chemically defined formulations.”

Developers who can use either type of media formulation retain all their options. But some companies insist as a risk-mitigating measure that all media be chemically defined. Those individuals, Dr. Dodson says, are “strapped” and may need to screen many formulations whose chemical makeup varies significantly.

BD Biosciences has designed and formulated BD Resurge™ CD1-CD5, a set of chemically defined, animal-free, and protein-free cell culture supplements. These graphs show the performance of the supplements in multiple base media across two CHO cell lines that express monoclonal antibodies at different baseline titers.

The Trisulfide Conundrum

Of the three main CHO cell culture parameters—dissolved oxygen (DO), temperature, and pH—DO is the most difficult to control. Variations in DO are common, and they may intensify through aeration strategies accompanied by high mass transfer rate. The result is sometimes a lactogenic culture and reduced product titers.

Higher lactate levels also correlate with an increase in the appearance of trisulfides in place of the more traditional disulfide linkages that hold together the heavy and light chains of therapeutic monoclonal antibodies.

The manufacture of antibody-drug conjugates (ADCs) requires reducing the sulfur-sulfur disulfide linkages to two thiol (–SH) groups. Drugs are subsequently attached to the antibody chains through a maleimide moiety. Elevated trisulfides therefore result in lower drug-antibody ratios (DARs) due to the presence of fewer sulfhydryl groups, and increased inhomogeneity among DAR species.

A Biogen group has reported variability in the prevalence of trisulfide linkages ranging from less than 1% to 39%. They found that feed strategies, particularly cysteine concentration, contributed to trisulfide formation due to the donation of hydrogen sulfide to the culture.

Meanwhile, Genentech scientists discovered that achieving stoichiometric conversion of antibodies to ADCs required adding the reducing agent tris(2-carboxyethyl)phosphine (TCEP) in quantities significantly above stoichiometric. Where two sulfhydryl groups form from one molecule of TCEP for disulfides, two equivalents of reagent were required to reduce the trisulfide.

Genentech scientists discussed the impact of trisulfide modification in a 2013 article that appeared in Bioconjugate Chemistry: “Antibodies with higher levels of trisulfide bonds require a greater TCEP:antibody molar ratio to achieve the targeted drug-to-antibody ratio.”

Michael Hippach, who leads a cell culture group at Agensys, worked at Biogen when his then-colleagues were beginning to understand the significance of trisulfides in ADC manufacture. At Agensys, Hippach conducted experiments to determine what other forces, if any, were responsible for lactogenicity. He found that the phenomenon is triggered by low DO during cell culture, and that lactate was associated with trisulfide formation, and therefore with product quality.

According to Hippach, trisulfides are a normal part of antibody production: “The analytical people say that trisulfide levels of less than 5% provide high-quality product.” Large companies such as Biogen can titrate for trisulfide reduction, but smaller firms typically lack the resources to do so. “We cannot conduct TCEP titration as much as we would like,” Hippach comments. “Trisulfides therefore become a quality issue.”

Drug-antibody ratios range up to eight, depending on the number of disulfide bridges in a molecule. ADC manufacturers often have a target ratio but cannot easily achieve it when trisulfide formation is not tightly controlled. Hippach says he was able to reach his target ratio when trisulfides remained below around 2%, but failed to achieve it once trisulfide levels rose above that.

Most cultures withstand DO fluctuations without a drop in quality, but the undesirable progression from high DO to lactogenesis, trisulfide formation, and reduced drug-antibody ratio pointed to Hippach’s solution, which is tighter control of DO.

“The problem is cells taking up too much oxygen,” he declares. Oxygen takeup was less easy to control with spargers that deliver high kLa but more manageable with a simple drill hole sparger. “When DO was very well controlled,” Hippach reports, “the trisulfide problem went away.”

Designing-in Purification

Designing or engineering cells to facilitate purification remains a goal of bioprocessors, says Günter Jagschies, Ph.D., strategic customer relations lead, GE Healthcare. “Most expression systems now provide titers that are acceptable in terms of economics and worldwide demand,” says Dr. Jagschies. “What we still do not understand well enough is how certain impurity profiles come about, and how those can be improved.”

Multiple (random) expression sites result in high titers but sometimes unacceptable product heterogeneity, while single-site expression controls impurities but lacks the productivity of multisite transfection. At the recent AccBio 2015 conference, for example, Regeneron presented on their EESYR® system for regulated high expression from a single inserted gene copy. The tradeoff to be made in yield for quality and ease of purification does not reach unacceptable levels, but may still be considered a step backwards, given the productivity requirements of regular production for a large market.

“The goal is to combine the robustness achievable with single-site expression with the productivity of multisite expression—without compromising quality,” insists Dr. Jagschies. “I believe this problem will be solved in a few years.” Solutions are sought more urgently with the introduction of complex therapeutic proteins that lack the complete IgG structure. For example, the components of bispecific monoclonal antibodies may combine in ways that increase the number of potential impurities through, say, incorrect assembly.

These issues may be resolved downstream, but overcoming them at the cell level, to produce only the desired molecule, is preferable. Cell-level investigations, however, are often considered impractical, given the desire to reduce time-to-clinic.

“Companies will not afford themselves the time to carry out this type of study,” Dr. Jagschies observes. “The current degree of robustness is ‘good enough,’ at least for preclinical studies.”
Bioprocessors will screen against cells with a tendency to generate aggregates, but do not usually consider examining the impurity profile exhaustively. Similarly, for all the talk about glycosylation, our understanding of this phenomenon—from the perspectives of both manufacturing and safety—is incomplete. “Again, the ‘good-enough’ approach,” Jagschies remarks.

This column has presented the growing trend toward continuous processing loudly and often. According to Dr. Jagschies, while most companies are looking into it, continuous processing is still considered quite novel and not ideal for every process.

The use of the term “continuous” stretches from “connecting several steps” (each operated in batch mode but running product through them without intermediate hold all the way) to steady-state perfusion culture. “I don’t see a clear trend yet, but I believe the former approach will be a first step to wider acceptance for truly continuous processing,”speculates Dr. Jagschies.

Currently, there are also business-related limitations. According to Dr. Jagschies, many CMOs would not be in a position to receive a continuous process, and a transfer of such a process might turn out to be difficult.

Dr. Jagschies, however, is much more certain about the industry’s embrace of high-density cell cultures. It should, he feels, lead to several productivity improvements. The idea is to inoculate the production culture from a high-density culture so cells can begin producing right away, or to segregate the seed train from the production culture, achieving higher scheduling flexibility in the plant. Productivity and quality improvements should apply to most of today’s production systems, not just mammalian cell cultures.

Microbioreactors’ New App

Microbioreactors are normally used to scale down or optimize batch and fed-batch processes. A notable example is the ambr microbioreactor, which was developed by TAP Biosystems, a company that is now part of Sartorius Stedim Biotech. Using ambr, scientists at Genzyme’s Geel, Belgium facility have developed a scale-down model for a long-term microcarrier-based perfusion process. Marijke Wynants, Ph.D., a process engineer for cell culture at Genzyme, says her group’s achievement is a first in the ever-expanding application base for microbioreactors.

“Some minor modifications were needed to establish a model for a perfusion process with ambr,” Dr. Wynants explains. “These were mostly software modifications to efficiently perform the protocols.” Genzyme uses the modified ambr system for process optimization and media screening. “The results are comparable to Genzyme’s at-scale processes and other currently used scale-down models.”

Microbioreactors, Dr. Wynants continues, require a new way of working: “A large number of samples and large quantity of data are generated in a short period of time, so efficient data processing and high-throughput downstream assays and equipment are needed.” The principal disadvantage is the existence of low sample volumes at times insufficient for determining product quality. When these are the circumstances, a larger scale-down model might still be needed for confirmation of results obtained by screening with ambr.

Dr. Wynants’ team found that there are times when usage of microcarriers in ambr can be challenging. “Rotation speed might need adjustment to create an environment in which cells can easily adhere to microcarriers,” notes Dr. Wynants. “Since microcarriers can be sticky, pipetting might need optimization, and possible coverage of DO and pH sensors with microcarriers might be an issue.”