January 15, 2011 (Vol. 31, No. 2)
New Approaches Facilitate Rapid, Low-Cost Analysis and Prediction of Product Yield
Cell culture media are complex products that are inherently difficult to analyze. Yet, media analysis is critical for assuring product quality. A group at National University of Ireland-Galway led by Alan Ryder, Ph.D., is investigating rapid, “holistic” fluorescence analysis for screening commercially used cell culture media.
“Holistic means taking on overall measurement of the media, rather than going down the separations route and identifying or quantifying every component,” Dr. Ryder explains.
His group has been working with Bristol-Myers Squibb (BMS) and other leading biomanufacturers on the technique, which is designed for Chinese hamster ovary (CHO) cell cultures. Eventually, it will become available to all biotech companies.
The technique uses readily available fluorescence instrumentation, costing about $20,000, and MATLAB, a fourth-generation engineering software package from MathWorks (www.mathworks.com). It further relies on chemometric analysis such as multiway robust principal component analysis (MROBPCA), n-way partial least-squares discriminant analysis, and regression (NPLS-DA).
MROBPCA assesses data variance, determines its statistical significance, and seeks to understand the source of the spectral variation. “If the variation is significant we use NPLS-DA to build identification models to discriminate different media based on their spectra, or NPLS to quantify the relationship between spectral data variation and production data, for example yield.”
This approach was developed empirically, by correlating samples of cell culture media from BMS with productivity measurements of manufacturing batches employing those specific media.
These sophisticated statistical tools are required since, while it identifies medium components through their natural fluorescence, this approach does not quantify ingredients individually. Rather, it provides a fluorescence map or signature that quantifies media quality and predicts its success. “We look at overall changes, not specific components.” Dr. Ryder claims the technique predicts volumetric productivity to within 0.13 g/L.
Chromatographic analysis of media components is possible but extremely time-consuming and expensive. Dr. Ryder’s rapid, low-cost technique is suitable for screening media rapidly and then, if needed, analyzing only select formulations later by HPLC for quality or trouble-shooting purposes.
The ability to observe compositional changes in media and predict product yield before production has enormous potential, Dr. Ryder says. “It should effectively eliminate one of the major process variables, thus leading to more consistent product quality and yield.”
When to Optimize?
CHO lines have become the workhorse cells for the production of monoclonal antibodies—arguably the most successful class of biologic drug. The stakes for efficient CHO media development are, therefore, high but the trend toward chemically defined media burdens companies with multiple mAb products. A one-size-fits-all approach to media formulation ignores significant phenotypic differences between cell lines, even those arising from a common parental cell, while creating unique media for specific clones is time-consuming and costly.
Genentech has solved this dilemma through creation of a platform medium that serves all its CHO-based products from early development through Phase II testing. If necessary, cell-line-specific media optimization occurs once a molecule enters late-stage development. This approach saves time and effort during the riskiest stages of drug development, while applying extensive quality-by-design science to late-stage compounds.
Feng Li, Ph.D., pilot plant manager at Genentech, explains that extensive differences among clones warrants this approach. “Clones can be remarkably different, even when they arise from the same parental cell lines, and even if you transfect with the same gene.” Some cells grow rapidly, others slowly, and some show markedly different metabolism, for example lactate production. “Gene integration during transfection is a random process.”
To develop a platform media that can accommodate different clones’ needs, Dr. Li’s group tested multiple model cell lines with various growth and metabolic properties to balance nutrient supplement.
Genentech often divides projects into early and late phases based on clinical stage. “Only during late-stage development when high titer is required, do we then conduct intensive media optimization,” Dr. Li says. “Due to the high attrition rate during biopharmaceutical development, it is important to have a good platform medium to fill the gap between early- and late-stage development.”
Having a platform medium, in which cells are expected to grow and express proteins well, if not robustly, greatly aids in clone selection and early-stage process development. And while it is true that optimization before clone selection might uncover some pleasant productivity surprises, “no company can afford individual medium optimization, at this stage, for so many clones.”
Dr. Li conducts optimization the old fashioned way, by applying a design-of-experiment approach to key media ingredients but mainly though trial and error. Two or three iterations are usually sufficient to achieve optimization. “If you have a decent platform medium and well-behaved cell lines the process can last from three to six months,” Dr. Li says, depending on the target expression level, basal performance, and project timelines. “But each cell line is different.”
Specialty Media and Ingredients
Culture media to support human embryonic stem cell (hESC) research and therapy is one of the hottest topics in biotech. Media must be exquisitely tuned to maintain cells in their desired stage, or to promote their proliferation or differentiation as desired.
A good deal of hESC work was carried out in murine-conditioned media that contained components of mouse embryonic fibroblasts (MEF). The use of murine ingredients to support hESCs has been “imperative” for the evolution of stem cell work, says David C. Hay, Ph.D., principal investigator at the MRC Centre for Regenerative Medicine in Edinburgh, Scotland.
However, murine-conditioned stem cell media suffer from many of the same limitations as serum-based media in biomanufacturing. “These include a general lack of definition, batch-to-batch variation, labor-intensive production, and their xenobiotic nature.”
According to Dr. Hay, fully characterized and defined media will be essential for hESC culture to progress to cGMP quality for therapeutic applications. Taking two significant steps in this direction, Dr. Hay’s group has begun standardizing hESC culture with two serum-free media: mTeSR (Stem Cell Technologies) and STEMPRO® (Life Technologies). Both media are serum-free and promote maintenance and differentiation of hESCs. Until now, stem cells were propagated in serum-free media but differentiated and maintained in conditioned (serum-containing) media.
“We’ve taken an incremental step forward toward defining and understanding culture conditions required for large-scale manufacture,” Dr. Hay reports.
Initial results have been encouraging. Either medium maintains cells in a pluripotent (but undifferentiated state) for more than 30 passages, and produces volumes of human hepatic tissue equal to those generated from cells maintained and propagated in conditioned media.
Dr. Hay is also looking to incorporate CELLstart™ (Life Technologies), a completely humanized defined substrate, in his stem cell work. CELLstart is meant to be used with STEMPRO as a substrate for stem cell attachment under serum-free conditions.
Protein hydrolysates have become standard cell culture media additives, replacing serum as the “magic,” undefined ingredient in many cell cultures. Hydrolysates, which are produced from yeast, soy, and other nonanimal sources, are complex products containing numerous components in sometimes widely disparate concentrations. The principal hydrolysate components are peptides and amino acids, but they also contain fatty acids, trace elements, vitamins, and carbohydrates.
“Benefits of hydrolysates can vary dramatically based on the raw material source, choice of hydrolytic enzyme, and process parameters,” says Christopher Wilcox, Ph.D., R&D director at Sheffield Bio-Science.
“Enhancement to cell growth and productivity is subject to the additive effect of hydrolysate components native to the basal medium, and those added through the hydrolysate,” Dr. Wilcox says. “You may not see the full benefit of hydrolysates unless you do up-front work to determine whether you’re duplicating ingredients or overdosing.”
Since cells thrive on unique concentrations of media ingredients, processors must fully understand, before they turn to hydrolysates, if their serum-free basal media already contains components that will be introduced through hydrolysates as well.
Dr. Wilcox views hydrolysates not as another source, like serum, of inconsistency and unpredictability, but as a vital step for the industry as a whole in its quest for 100% defined media. He notes that within a particular class of hydrolysate Sheffield can provide numerous variants depending on the extent or chemical specificity of the hydrolysis, removal of certain components, and other factors.
Improving Quality and Safety
Of the several significant microbial contaminants that plague mammalian cell culture manufacturing, bacteria are more common but viruses are more devastating. Adventitious viral infections—as opposed to viruses emerging from the genomes of cultured cells—always lead to loss of product and often to extended plant shutdowns.
No company is immune from viruses, no matter how rigorous their sourcing and cleaning practices. Genzyme, Genentech, and Amgen have all been affected, with the minute virus of mice (MVM) particularly troublesome.
“The only way to deal with a viral infection is to quarantine the entire building, throw away everything you possibly can, shrinkwrap the facility, and fumigate,” says Bob Weaver, senior scientist at Amgen.
Raw materials, including media ingredients, are a high-risk entry point for viral contamination. Today’s preference for better defined, animal component-free media requires 50 or more discrete media ingredients, each one multiplying the risk of MVM or other viruses entering the production cycle. “One virus particle can contaminate an entire plant,” Weaver observes.
The solution that Amgen has adopted is risk mitigation at point of entry. Ingredients are treated with one of two virucidal technologies borrowed from other industries: ultraviolet-C (UV-C) and high-temperature short-time pasteurization (HTST). Widely used in the sterilization of water, beverages, and blood plasma, short-wave UV-C includes the germicidal 253.7 nm wavelength. HTST is widely used in the food industry and is similar to other types of heat treatments used in biotech.
Genentech applies these treatments after formulation, which assures that the media is sterile at the point of use. According to Weaver every large mammalian cell culture facility is either using or considering deployment of media sterilization. “Although it’s expensive we’re considering it for all our products. If you’re in this business long enough, chances are that sooner or later your process will be exposed to some type of virus.”
Regulators have recently become concerned with media components that are not cleared by purification unit operations. The targets are not amino acids and nutrients, but additives such as anti-foaming agents and detergents. “These substances are not listed in the drug substance specification, as purification steps are expected to remove them,” notes Suzanne Aldington, Ph.D., a group leader at Lonza Biologics. But chromatography and membrane filtration do not always do the job adequately.
While these substances are not considered unsafe at the levels at which they are added, regulators are nevertheless concerned because their use and carry-over through downstream processing has not been subjected to a thorough risk-assessment study.
Chromatography, buffer exchange, membrane adsorption, and filtration are known to remove most of these materials but no-one knows to what extent they do. For example, detergents should wash through a protein A column, ultrafiltration membranes trap impurities below 10 kDa, and antifoaming agents stick to filters.
Still, Dr. Aldington’s group is looking into assays that will quantify additives at each purification step. These may include spiking studies where known quantities of impurity are loaded, at small scale, onto a column and the eluent tested by HPLC.
It is unknown whether FDA and its European counterpart, EMA, will require specific assays for common media and process additives, but Dr. Aldington suggests that the likelihood of having to add specific purification steps to clear trace additives is not very great.