Controlling Glycosylation for Improved Product Quality

nicolas_/Getty Images

The majority of top-selling biologics and approximately half of the approved therapeutic proteins contain glycosyl structural components, according to Mark Rogers, vice-president USA-life science with SGS. These sugar modifications play important roles in the function, stability, and potency of biologic drugs. Glycosylation is, therefore, a critical quality attribute (CQA) of many biopharmaceuticals. Until recently, control of glycosylation during cell culture and fermentation has been challenging.

Greater understanding of metabolic pathways has allowed more effective process optimization, which has also been facilitated by the development of high-throughput analytical methods. In some cases, engineering of cell lines has enabled the use of cells that exhibit desirable glycosylation behavior. More recently, in-vitro glycoengineering has been introduced as a method for moving control of glycosylation from the complex conditions in the bioreactor to a simpler downstream processing environment.

Why the need for control?
Variation in glycosylation can impact the safety, efficacy, and clearance of a therapeutic product. Therefore, control of glycosylation patterns is essential in biopharmaceutical development and manufacturing.

“Glycosylation has the potential to make or break the successful market entry of a drug,” states Roland Dorn, international product manager, CustomBiotech, Roche Diagnostics. As the most frequent form of post-translational modification, glycosylation involves the attachment of sugars onto the backbone of therapeutic proteins.

The presence of different sugars at different positions can affect protein folding, stability, and activity, such as antibody dependent cellular cytotoxicity (ADCC) and general pharmacokinetics. For instance, it is known that sialylation of erythropoietin (EPO) is important for good pharmacokinetic properties.

Glycosylation during fermentation naturally occurs with a certain degree of heterogeneity and can be affected by many different factors, such as the expression system, process conditions, or media composition and feed protocols. Consequently, there is a certain risk that the glycosylation pattern can vary from batch to batch, leading to variations in product quality. Not surprisingly, therefore, glycosylation is considered to be a CQA that must be controlled to ensure good control of protein quality.

Role of metabolic studies and modeling
Working through the metabolic pathways for post-translational modifications done to proteins in cells has allowed companies to identify the correct nutrients that help to modify the end product quality parameters, according to Michelle Yannetti, director of market development for pharma analytics and purification with Thermo Fisher Scientific.

To support a quality-by-design (QbD) approach to the optimization of glycosylation processes, several mathematical models have also been developed that combine critical elements of glycan production, such as the N-linked glycosylation process in the Golgi, cell-culture dynamics, and nucleotide sugar biosynthesis. “It is hoped that, in the future, such models will result in the optimization of bioprocess design and control for the desired glycan features without the need for large experimental programs,” Rogers adds. Yannetti also sees growing interest, particularly given the large quantity of available data, to develop models that can predict needed process and nutrient patterns even more quickly than the methods used today.

Impact of improved analytical capabilities
Early efforts to consider product quality, particularly glycosylation, involved long lead times for processing of samples and then running appropriate chromatography and mass spectrometry (MS) assays to determine the glycoform distribution for a target protein. “This issue often restricted protein quality testing to initial clone output and then at the end of process development, leaving more to chance and often the observation shifts from initial output to final process with little known about the causes of the changes,” explains Yannetti.

The ability to monitor protein glycosylation has improved dramatically even in the past five years, according to Byron Kneller, director of analytical and formulation development at CMC Biologics. “For some molecules, intact protein liquid chromatography-mass spectrometry (LC-MS), either with or without a one-step protein purification, gives adequate glycoform resolution to enable process decisions to be made,” he notes. In addition, modern high-resolution MS systems (quadrupole-time-of-flight, Orbitrap) provide sufficient resolution for protein glycoforms provided the heterogeneity is not too great.

Fortunately, Kneller adds, in cases where either glycoform heterogeneity is too high or more detail is needed, the throughput of released glycan methods has improved sufficiently to permit analysis of large numbers of samples in a reasonable timeframe. Methods have become faster for sample preparation, and there are even highly sensitive, high-throughput alternatives available today that can allow analysis of more samples, thus enabling the monitoring of glycan changes throughout process development, agrees Yannetti.

For monoclonal antibodies (mAbs), one routine method is ultra high-pressure liquid chromatography (UHPLC) of fluorescent-tagged glycans after enzymatic release. Such techniques have had significant application in the biosimilar field, according to Rogers, where much emphasis is placed on the ability to detect minor differences in glycan structures. Thermo Fisher Scientific, for example, has developed proprietary dyes that can be used to easily detect labeled glycans in low femtomolar concentrations, and is developing additional dyes with even greater sensitivities.

Rogers agrees that mass spectrometry has become the method of choice for many applications. “This technique is able to provide characterization data regardless of the glycan type, is inherently sensitive, and can be adapted for routine release testing,” he comments. It is also one of the few methodologies that is able to reveal simultaneous information pertaining to both the protein and glycan moieties and allows data to be obtained without the need for structural modification.

A triage approach can be effective in some cases, according to Kneller. With this approach, a rapid but lower-resolution method is used to identify a small number of potentially promising conditions or clones from many, and then this small number of samples is reanalyzed with a higher-resolution method to enable more confident ranking.

Advances in cell-line engineering and bioprocess design
Considerable improvements have been made in the development and production of biologics, particularly in the field of mAbs, according to Rogers. “The glycoforms of these particular therapeutics have been refined from early, relatively diverse glycan populations to simple, homogenous structures with very little batch-to batch variability due to increased understanding of the controllable bioprocess parameters that affect glycosylation, such as supplement concentration, pH, dissolved oxygen, cell-line selection, and temperature,” he explains.

Glycosylation inhibitors have also been effectively used during production to manipulate the final glycosylated product. For example, in the case of the mannosidase II inhibitor swainsonine, glycoproteins with hybrid glycan structures may be selectively produced.

At CMC Biologics, sophisticated clone screening, media optimization, and the use of sophisticated scale-down models during process development lead to large-scale processes that are highly robust when it comes to glycosylation, according to Mark Rice, CMC Biologics’ director of cell line and upstream process development. “It is much easier to pick a clone from a parental cell-line with favorable glycopatterns than to ‘fix’ a clone using basal and feed media, supplements, hydrolysate additions, and process changes,” he notes. In addition, under the latter circumstances, there is typically an unpleasant trade-off between the CQA and titer. “Engineering of cell-lines to produce specific glycan patterns has great promise and also offers an ‘out-of-the-box’ solution, potentially saving a lot of development time,” he observes.

Yannetti, on the other hand, sees things differently. “Although several groups have demonstrated the ability to engineer cells to eliminate undesirable glycoforms (for example engineered a-fucosylation to enhance ADCC activity), the recent advances in understanding of metabolic pathways and the impact of nutrients and process shift parameters appear to provide the greatest advances in this area, leading to greater control and consistency in this aspect of protein quality,” she says. “Today, with the knowledge we have gained in delivery of nutrients to culture, it is possible to provide predictable glycosylation targeting and to resolve changes in glycoforms that are not desirable with both process and nutrient inputs.”

She points to a new feed system from Thermo Fischer Scientific that can provide predictability by simply changing to the new feed at different times in the process, thus allowing the selection of feed and process parameters that lead to greater consistency of glycosylation output from a culture process.

In-vitro glycoengineering looks promising
Scientists at Roche Diagnostics have found a way to uncouple glycosylation management from complex cell-culture and fermentation processes. The Roche CustomBiotech portfolio of well-characterized enzymes and activated sugars for in-vitro glycoengineering (IVGE) is designed for specific and efficient alteration of sugar moieties during downstream processing. The animal- and antibiotic-free glycosyltransferases in the portfolio have demonstrated high activity on a variety of glycoproteins. They can be made available in gram to kilogram amounts and in compliance with GMP requirements, if needed, according to Dorn.

“Isolating glycosylation management from fermentation separates strategies to optimize glycosylation and yield, granting greater control over each,” he asserts. With IVGE, he adds that certain glycoforms (e.g., galactosylation and sialylation) can be enriched in downstream processing using discrete enzymatic reactions with clear kinetics and predictable outcomes, thus avoiding the need for uncertain tweaking of bioprocesses and leading to potentially shorter development times and improved control of the manufacturing process.

Extensive studies were conducted to demonstrate the performance of the different enzymes. The results confirmed that they are selective for N-glycosylation of a number of different antibodies and other target proteins, according to Dorn. Reactions can be run in different buffers, but the ionic strength of the buffer should be considered because it can influence the behavior of the enzymes. Incubation at 37 °C leads to completion of reactions in typically 24 hours or less. The optimal concentration of the enzyme is dependent on the target protein.

In addition, galactosylation followed by sialylation can be performed in a one-pot reaction without purification of the protein after the first step, but the degree of double sialylation is slightly lower than when the two reactions are performed separately, according to Dorn. Other studies demonstrated that performing the reaction on the eluate from

Protein A capture chromatography provides the simplest way to incorporate IVGE into the downstream purification process. Roche is currently developing a residual enzyme assay that will allow for confirmation that no glycosyl transferases are present in the protein product.

Finally, it was demonstrated that IVGE can be used to study the correlation between glycan profiles and receptor binding behaviors. Currently, Dorn also notes that there are no indications that IVGE treatment would lead to any substantial changes in the microheterogeneity of proteins.

More work needed, but progress significant
“Ten years ago, we were happy to shift the charge distribution on a gel. Today, because the sophistication of our understanding of glycoforms and their correlation with observed pharmacokinetics (PK) has grown dramatically, our standard is to maximize specific glycan forms that have known correlations to good PK and/or are observed in vivo,” asserts Rice. He adds that increased throughput due to better equipment and analysis for clone evaluation has enabled earlier and better screening and avoidance of problematic cell lines and the need to fix glycosylation issues later in the process.

This knowledge is largely related to N-linked glycans of mAbs, given that the focus of bioprocess optimization and control for glycosylation has been with this successful class of biologic APIs. Other types of glycosylation, such as O-linked and C-linked structures, are also of significance in certain potential therapeutic biologics, notes Rogers, but the biosynthetic pathways are quite distinct from those producing the N-linked structures and have not yet been subject to the same investigative scrutiny in terms of bioprocess control. He thus believes that greater understanding and control of the glycosylation processes in the future is likely to impact the production of other therapeutics such as viral vaccines, where the outer coat of glycoproteins can have a marked impact on immunogenicity mediated via the glycan structures.

The greatest need today, according to Yannetti, is for innovator companies to determine which glycan forms are most desirable for proteins in different therapeutic applications. “Once there are clear targets, it may be possible to further refine options to control glycan profiles from the start of development, and even possibly at the level of stable clone selection and development to fit with platform processes,” she says.

Rogers, meanwhile, expects that as the understanding of specific glycoform structures and functions improves, it is likely that future techniques will allow a single desired glycoform to be isolated or produced, either by cell line selection and/or the application of specific process controls. “Of course,” he observes, “protein engineering will continue to allow manipulation of the number and position of glycosylation sites, resulting in enhancement of certain properties such as potency and half-life, as already achieved with EPO.”

Article DetailsBioPharm International
Vol. 29, No. 6
Pages: 32–34, 40

Citation
When referring to this article, please cite it as C. Challener, "Controlling Glycosylation for Improved Product Quality" BioPharm International 29 (6) 2016.